In this paper, we study the usage of machine learning models for sales time series forecasting. The effect of machine learning generalization has been considered. A stacking approach for building regression ensemble of single models has been studied. The results show that using stacking technics, we can improve the performance of predictive models for sales time series forecasting.

In recent years, transformer-based models have gained prominence in multivariate long-term time series forecasting (LTSF), demonstrating significant advancements despite facing challenges such as high computational demands, difficulty in capturing temporal dynamics, and managing long-term dependencies. The emergence of LTSF-Linear, with its straightforward linear architecture, has notably outperformed transformer-based counterparts, prompting a reevaluation of the transformer's utility in time series forecasting. In response, this paper presents an adaptation of a recent architecture termed extended LSTM (xLSTM) for LTSF. xLSTM incorporates exponential gating and a revised memory structure with higher capacity that has good potential for LTSF. Our adopted architecture for LTSF termed as xLSTMTime surpasses current approaches. We compare xLSTMTime's performance against various state-of-the-art models across multiple real-world datasets, demonstrating superior forecasting capabilities. Our findings suggest that refined recurrent architectures can offer competitive alternatives to transformer-based models in LTSF tasks, potentially redefining the landscape of time series forecasting.

Modern energy systems collect high volumes of data that can provide valuable information about energy consumption. Electric companies can now use historical data to make informed decisions on energy production by forecasting the expected demand. Many deep learning models have been proposed to deal with these type of time series forecasting problems. Deep neural networks, such as recurrent or convolutional, can automatically capture complex patterns in time series data and provide accurate predictions. In particular, Temporal Convolutional Networks (TCN) are a specialised architecture that has advantages over recurrent networks for forecasting tasks. TCNs are able to extract long-term patterns using dilated causal convolutions and residual blocks, and can also be more efficient in terms of computation time. In this work, we propose a TCN-based deep learning model to improve the predictive performance in energy demand forecasting. Two energy-related time series with data from Spain have been studied: the national electric demand, and the power demand at charging stations for electric vehicles. An extensive experimental study has been conducted, involving more than 1900 models with different architectures and parametrisations. The TCN proposal outperforms the forecasting accuracy of Long Short-Term Memory (LSTM) recurrent networks, which are considered the state-of-the-art in the field.

Time series forecasting has been a challenging area in the field of Artificial Intel-ligence. Various approaches such as linear Neural Networks, Recurrent Linear Neural Networks, Convolutional Neural Networks and recently Transformers have been attempted for the time series forecasting domain. Although, Trans-former-based architectures have been outstanding in the Natural Language Pro-cessing domain, especially in autoregressive language modeling, the initial at-tempts to use transformers in the time series arena have met with mixed success. One of the recent papers demonstrated that modeling the multi-variate time se-ries problem as independent channels on a transformer architecture produced lower mean square and mean absolute errors on benchmarks, however, another recent paper claimed that a simple linear neural network out-performs previous approaches including transformer-based designs. We investigate this paradox in detail comparing the linear Neural Network and Transformer based designs, provide insights to why a certain approach may be better for a particular type of problem. We also improve upon the recently proposed simple linear Neural Network based architecture by using dual pipelines with batch normalization and reversible instance normalization. Our enhanced architecture outperforms all existing architectures for time series forecasting on majority of the popular benchmarks.

The environmental issues we are currently facing require long-term prospective efforts for sustainable growth. Renewable energy sources seem to be one of the most practical and efficient alternatives in this regard. Understanding a nation's pattern of energy use and renewable energy production is crucial for developing strategic plans. No previous study has been performed to explore the dynamics of power consumption with the change in renewable energy production on a country-wide scale. In contrast, a number of deep learning algorithms demonstrated acceptable performance while handling sequential data in the era of data-driven predictions. In this study, we developed a scheme to investigate and predict total power consumption and renewable energy production time series for eleven years of data using a Recurrent Neural Network (RNN). The dynamics of the interaction between the total annual power consumption and renewable energy production are investigated through extensive Exploratory Data Analysis (EDA) and a feature engineering framework. The performance of the model is found satisfactory through the comparison of the predicted data with the observed data, visualization of the distribution of the errors and Root Mean Squared Error (RMSE) value of 0.084. Higher performance is achieved through the increase in the number of epochs and hyperparameter tuning. The proposed framework can be used and transferred to investigate the trend of renewable energy production and power consumption and predict the future scenarios for different communities. Incorporation of the cloud-based platform into the proposed pipeline may lead to real-time forecasting.

Rapid growth and domestic migration to capital city, Kathmandu Metropolitan City (KMC), Nepal have amplified land-use transition to unmanaged and unplanned urbanization have exacerbated problems of Urban Heat Island (UHI) Effect. Increasing built up surfaces, means of transport and industrial activities are major results for increasing temperature in the city area as compared to other areas. This study focuses on forecasting Land Surface Temperature (LST) using both univariate and multivariate time series models. The primary goal is to predict the impact of climatic factors on the Urban Heat Island (UHI) effect, utilizing data from 1981 to 2019. We analyzed key factors including - (i) relative humidity, (ii) surface pressure, (iii) specific humidity, and (iv) wind speed sourced from the MEERA-2 dataset. A univariate model using exponential smoothing achieved a Mean Absolute Error (MAE) of 3.17°C and an R² score of 0.94, highlighting strong predictive accuracy. Additionally, the multivariate Prophet model, which incorporates trend and seasonal components, further refined the forecasting by integrating multiple climatic factors. This study aims provides valuable information via prediction for sustainable urban planning and landscape design in KMC and other 61 districts.

Today, there is a time series analysis that has gained significant popularity across various industries to provide effective solutions for any specific challenges. Artificial intelligence logic is generally used to solve time-series problems with the help of various frameworks that are available for the development of AI-based applications. Currently, there is a significant framework in Python that is driven by its comprehensive libraries and their practical implementations. Time series analysis can be built with the help of some specific models that give results to differentiate the efficiency and domains of great work. This paper focuses on three diverse libraries of Python to predict the Gulf stock exchange market. These libraries are prediction models that are helpful in predicting future values. These models utilized include SARIMA (Seasonal Auto-Regressive Integrated Moving Average), Neural Prophet Model, and Fb Prophet model. Each model is trained using the provided datasets, which are followed by evaluation parameters such as MSE, RMSE, recall, MDAPE, accuracy, precision, etc. The main purpose of using these three models for stock market prediction is to find the better-suited models. In the past few decades, the stock market industry has seemed to have promising growth after 2022 onwards. This paper will help us make a significant decision about model usage for addressing such complex challenges. Especially, the research findings indicate that all three models demonstrate in different domains. SARIMA can perform well in overall prediction tasks, while the Prophet model works great with smaller datasets that contain seasonality trends.

To improve the monitoring of the electrical power grid it is necessary to evaluate the influence of contamination in relation to leakage current and its progression to a disruptive discharge. In this paper, insulators were tested in a saline chamber to simulate the increase of salt contamination on their surface, and to evaluate the supportability of these components. From the time series forecasting of the leakage current, it is possible to evaluate the development of the fault before a flashover occurs. Choosing which method to use is always a difficult task since some models may have higher computational effort. In this paper, for a complete evaluation, the long short-term memory (LSTM), group method of data handling (GMDH), adaptive neuro-fuzzy inference system (ANFIS), bootstrap aggregation (bagging), sequential learning (boosting), random subspace, and stacked generalization (stacking) ensemble learning models are analyzed. A review and comparison of these well-established methods for time series forecasting is performed. From the results of the best structure of the model, the hyperparameters are evaluated and the Wavelet transform is used to obtain an enhanced model.

One of the key components of application performance monitoring (APM) software is 2 AI/ML empowered data analytics for predictions, anomaly detection, event correlations and root 3 cause analysis. Time series metrics, logs and traces are three pillars of observability and the valuable 4 source of information for IT operations. Accurate, scalable and robust time series forecasting and 5 anomaly detection are desirable capabilities of the analytics. Approaches based on neural networks 6 (NN) and deep learning gain increasing popularity due to their flexibility and ability to tackle complex 7 non-linear problems. However, some of the disadvantages of NN-based models for distributed cloud 8 applications mitigate expectations and require specific approaches. We demonstrate how NN-models 9 pretrained on a global time series database can be applied to customer specific data using transfer 10 learning. In general, NN-models adequately operate only on stationary time series. Application 11 to non-stationary time series requires multilayer data processing including hypothesis testing for 12 data categorization, category specific transformations into stationary data, forecasting and backward 13 transformations. We present the mathematical background of this approach and discuss experimental 14 results from the productized implementation in Wavefront by VMware (an APM software) while 15 monitoring real customer cloud environments.

When dealing with limited sample sizes or missing values in time series data, achieving high accuracy in forecasting becomes challenging. This challenge is particularly common in wind energy forecasting, where many wind farms lack complete data. In this paper, we employ the widely-used ARIMA (Autoregressive Integrated Moving Average) model for time series forecasting. To address the difficulties posed by limited sample sizes and missing values, we adopt a unique approach by utilizing multiple ARIMA models for different datasets. A key distinction from other multiple models methods for time series forecasting lies in each model within our approach using its own dataset, making the fusion process more intricate. To overcome this, we incorporate neural networks and two techniques for model fusion: direct fusion with a learning mechanism and compensation for missing data through training. For each ARIMA model, we introduce a noise estimation method to circumvent the white noise assumptions inherent in ARIMA models. Finally, we apply these methods to wind energy prediction, and our experimental results showcase a significant enhancement in forecasting accuracy.

Due to its flexibility, cloud computing has become essential in modern operational schemes. However, the effective management of cloud resources to ensure cost-effectiveness and maintain high performance presents significant challenges. The pay-as-you-go pricing model, while convenient, can lead to escalated expenses and hinder long-term planning. Consequently, FinOps advocates proactive management strategies, with resource usage prediction emerging as a crucial optimization category. In this research, we introduce the multi-time series forecasting system (MSFS), a novel approach for data-driven resource optimization alongside the hybrid ensemble anomaly detection algorithm (HEADA). Our method prioritizes the concept-centric approach, focusing on factors such as prediction uncertainty, interpretability and domain-specific measures. Furthermore, we introduce the similarity-based time-series grouping (STG) method as a core component of MSFS for optimizing multi-time series forecasting, ensuring its scalability with the rapid growth of the cloud environment. The experiments performed demonstrate that our group-specific forecasting model (GSFM) approach enabled MSFS to achieve a significant cost reduction of up to 44%.

Accurate prediction of crop production is essential in effectively managing agricultural countries' food security and economic resilience. This study evaluates the performance of statistical and machine learning-based methods for large-scale crop production forecasting. We predict the quarterly production of 325 crops (including fruits, vegetables, cereals, non-food, and industrial crops) across 83 provinces in the Philippines. Using a comprehensive dataset of 10,949 time series over 13 years, we demonstrate that a global forecasting approach using a state-of-the-art deep learning architecture, the transformer, significantly outperforms traditional local forecasting approaches built on statistical and baseline methods. By leveraging cross-series information, our proposed way is scalable and works well even with time series that are short, sparse, intermittent, or exhibit structural breaks/regime shifts. The results of this study further advance the field of applied forecasting in agricultural production and provide a practical and effective decision-support tool for policymakers that oversee the farm sector on a national scale.

The evaluation of geometric defects is necessary in order to maintain the integrity of structures over time. These assessments are designed to detect damages of structures and ideally help inspectors to estimate the remaining life of structures. Current methodologies for monitoring structural systems, while providing useful information about the current state of a structure, are limited in the monitoring of defects over time and in linking them to predictive simulation. This paper presents a new approach to the predictive modeling of geometric defects. A combination of segmented from point clouds are parametrized using the convex hull algorithm to extract features from detected defects, and a stochastic dynamic model is then adapted to these features to model the evolution of the hull over time. Describing a defect in terms of its parameterized hull enables consistent temporal tracking for predictive purposes, while implicitly reducing data dimensionality and complexity as well. In this study, 2D point clouds analogous to information derived from point clouds were first generated over simulated life-cycles. The evolutions of point cloud hull parameterizations were modeled as stochastic dynamical processes via autoregressive integrated moving average (ARIMA) and vectorized autoregression (VAR) and compared against ground truth. The results indicate that this convex hull approach provides consistent and accurate representations of defect evolution across a range of defect topologies and is reasonably robust to noisy measurements, however assumptions regarding the underlying dynamical process play a significant the role in predictive accuracy. The results were then validated on experimental data from fatigue testing with high accuracy. Longer term, the results of this work will support finite element model updating for predictive analysis of structural capacity.

This paper addresses the evolving landscape of electricity markets in Europe, with a focus on the integration of Renewable Energy Communities as introduced by the Renewable Energy Directive 2018/2001. Residents within a postal code area are highly incentivized to join a community, which enables them to exchange energy among themselves at lower procurement costs. Thereby, energy management systems optimize the operation of respective energy systems, with electrical load forecasting playing a key role. Given that prosumers may switch between communities on a daily basis, electricity demands within these groups will vary, leading to data that is non-stationary, discontinuous as well as non-identical and independently distributed. To encounter this issue, we propose a sophisticated forecasting model that applies federated learning, using informations from various distributed communities to learn domain-invariant features. To achieve this, we initially utilize synthetic electrical load time series at district level and aggregate them to profiles of Renewable Energy Communities with dynamic portfolios. Subsequently, we develop a forecasting model that accounts for the composition of residents of a Renewable Energy Community, adapt data pre-processing in accordance with the time series process, and detail a federated learning algorithm that incorporates weight averaging and data sharing. Following the training of various experimental setups, we ultimately evaluate their effectiveness. The findings suggest that our proposed framework is capable of effectively forecast non-stationary and discontinuous time series and that it can be applied to new, unseen data through the integration of knowledge from multiple sources.

Continuous and uncontrolled extraction of groundwater often creates tremendous pressure on groundwater levels. As a part of the sustainable planning and effective management of water resources, it is crucial to assess the existing as well as future groundwater level (GWL) condition. In the current study, an attempt was made to model and forecast GWL using artificial neural networks (ANN) and multivariate time series models. Autoregressive integrated moving average (ARIMA) and ARIMA incorporating exogenous variables (ARIMAX) were adopted as the time series models. Kushtia district in Bangladesh was selected as the case study area, and GWL data of five monitoring wells in the study are used to demonstrate the modeling exercise. Rainfall was taken as the exogeneous variable to explore whether its inclusion enhanced the performance of GWL forecasting using the developed models. The performance of each time series and ANN model was assessed based on various model evaluation criteria. It was evident from the results that the multivariate ARIMAX model (SSE of 15.361) performed better than the univariate ARIMA model with an SSE of 17.217 for GWL forecasting. This demonstrates the fact that the multivariate time series models generated enhanced forecasting of GWL compared to the univariate time series models. When comparing the time series and ANN models, it was found that the ANN-based model outperformed the time series models with the enhanced forecasting accuracy (SSE of 9.894). Results also exhibit a significant correlation coefficient value R of 0.995 (ANN 6-8-1) for the existing and predicted data. The current study conclusively proves the superiority of ANN over the time series models for the enhanced forecasting of GWL in the study area. Thus, the ANN approach was not only carried out for model building and simulation but also to provide a valuable tool for managing water resources amidst changing environmental conditions.

Accurately predicting electrical signals of three-phase Direct Torque Control (DTC) induction motors is crucial for optimal motor performance and effective condition monitoring. However, multiple DTC motors’ complexity and operating conditions’ variations pose challenges for traditional prediction methods. To overcome these challenges, we propose an innovative approach that combines Fast Fourier Transform (FFT) to transform electrical motor simulation data and a Bi-directional Long Short-Term Memory (Bi-LSTM) network to forecast processed motor data. By transforming the DTC induction motor signals from the time domain to the frequency domain, we enhance the capability of learning models to capture subtle differences and generate various input features. The Bi-LSTM model effectively captures both forward and backward dependencies in time series data, enabling accurate prediction of electrical signals. We evaluate the proposed approach using our simulated dataset and compare the proposed method to a state-of-the-art model, the Gated Recurrent Unit (GRU), demonstrating its effectiveness in improving the accuracy and reliability of induction motor signals forecasting. The finding provides valuable insights for advancing motor control and operation in industrial applications.

This study presents a novel approach for predicting hierarchical short time series. In this article, our objective was to formulate long-term forecasts for household natural gas consumption by considering the hierarchical structure of territorial units within a country's administrative divisions. For this purpose, we utilized natural gas consumption data from Poland. The length of the time series was an important determinant of the data set. We contrast global techniques, which employ a uniform method across all time series, with local methods that fit a distinct method for each time series. Furthermore, we compare the conventional statistical approach with a machine learning (ML) approach. Based on our analyses, we devised forecasting methods for short time series that exhibit exceptional performance. We have demonstrated that global models provide better forecasts than local models. Among ML models, neural networks yielded the best results, with the MLP network achieving comparable performance to the LSTM network while requiring significantly less computational time.

Snow is a vital environmental parameter and dynamically responsive to climate change, particularly in mountainous regions. Snow cover can be monitored at variable spatial scales using Earth Observation (EO) data. Long-lasting remote sensing missions enable the generation of multi-decadal time series and thus the detection of long-term trends. However, there have been few attempts to use these to model future snow cover dynamics. In this study, we therefore explore the potential of such time series to forecast the Snow Line Elevation (SLE) in the European Alps. We generate monthly SLE time series from the entire Landsat archive (1985-2021) in 43 Alpine catchments. Positive long-term SLE change rates are detected, with the highest rates (5-8 m/y) in the Western and Central Alps. We utilize this SLE dataset to implement and evaluate seven uni-variate time series modeling and forecasting approaches. The best results were achieved by Random Forests, with a Nash-Sutcliffe efficiency (NSE) of 0.79 and a Mean Absolut Error (MAE) of 258 m, Telescope (0.76, 268 m), and seasonal ARIMA (0.75, 270 m). Since the model performance varies strongly with the input data, we developed a Combined forecast based on the best performing methods in each catchment. This approach was then used to forecast the SLE for the years 2022-2029. In the majority of the catchments the shift of the forecast median SLE level retained the sign of the long-term trend. In cases where a deviating SLE dynamic is forecast a discussion based on the unique properties of the catchment and past SLE dynamics is required. In the future, we expect major improvements in our SLE forecasting efforts by including external predictor variables in a multi-variate modeling approach.

The goal of VAR or BVAR is the characterization of the dynamics and endogenous relationships among time series. Also the VAR models are known for their applications to forecasting and policy analysis. This paper compare the performance of VAR and Sims-Zha Bayesian VAR models when the multiple time series are jointly influenced by different levels of collinearity and autocorrelation in the short term (T=16, 32, 64 and 128). Five levels (-0.9,-0.5, 0,+0.5,+0.9) of collinearity and autocorrelation were considered and the results from the simulation study revealed that VAR(2) model dominated for no and moderate levels of autocorrelation (-0.5, 0, +0.5) irrespective of the collinearity level except in few cases when T=16. While the BVAR models dominated for high autocorrelation levels (-0.9 and +0.9) irrespective of the collinearity level except in few cases when T=128. The performance of the models varies at different levels of the collinearity and autocorrelated error, and also varies with the short term periods. Furthermore, the values of the RMSE and MAE criteria decrease as a result of increase in the time series length. In conclusion, the performance of the forecasting models depend on the time series data structure and the time series length. It is therefore recommended that the data structure and series length should be considered in using an appropriate model for forecasting.

Time irreversibility, i.e. the lack of invariance of a system under the operation of time reversal, has long attracted the attention of the statistical physics community; and has been shown to be a relevant marker of altered dynamics in many real-world problems. We here introduce and analyse the complementary problem of its manipulation. In other words, we ask whether, given a time series, this can be manipulated to achieve a desired irreversibility while maintaining the original dynamics. We show how this problem can be tackled using Continuous Ordinal Patterns, a non-linear transformation of a time series based on the local structure created by neighbouring values. We further illustrate the relevance of this problem in the context of brain dynamics, obtaining that schizophrenic patients and control subjects are characterised by different "distances to irreversibility". We finally discuss some open questions, including the meaning of such manipulation from both theoretical and applied viewpoints.

The Covid-19 pandemic and the subsequent implementation of lockdown measures have significantly impacted global electricity consumption, necessitating accurate energy consumption forecasts for optimal energy generation and distribution during a pandemic. In this study, we propose a new forecasting model called the Multivariate Multilayered LSTM with Covid-19 case injection ($\proposedModel$) for improved energy forecast during the next occurrence of a similar pandemic. We utilize data from commercial buildings in Melbourne, Australia during the Covid-19 pandemic to predict energy consumption and evaluate the model's performance against commonly used methods such as LSTM, Bi-LSTM, Linear Regression, Support Vector Machine and the previously published work of Multilayered LSTM (M-LSTM). The proposed forecasting model was analyzed using the following metrics of mean percent absolute error (MPAE), normalized root mean square error (NRMSE), and $R^2$ score values. The model $\proposedModel$ demonstrates superior performance, achieving the lowest MPAE values of 0.061, 0.093, and 0.158 for data sets from 3 different buildings, respectively. Our results highlight the improved precision and accuracy of the model, providing valuable information for energy management and decision-making during the challenges posed by the occurrence of a pandemic like Covid-19 in the future.

The anomaly detection task is very important in computer science. And there are a lot of anomaly detection methods. Different from some thresholding methods, some unsupervised methods could make us get more accurate and faster result, which is the object of the project. In this paper, I tried to use EWMA and some other methods in two datasets: Webank time consuming indicators dataset and AIOps Challenge dataset. The paper consists nine parts: background of the project, related work, description of algorithms, implementation details, experimental setup and data sets used, experimental results and discussion, future directions, reference and meeting notes.

Dhaka, one of the fastest-growing megacities, is highly vulnerable to urban flooding due to rapid urbanization, inadequate drainage systems, and climate change-induced extreme weather events. This study describes the development of a comprehensive urban flood forecasting system for Dhaka City using the MIKE+ hydrodynamic model. The system integrates hydrological and hydraulic components, modeling complex interactions among surface runoff (pluvial), riverine overflow (fluvial), drainage networks, and overland flow dynamics. The model utilizes real-time data from river gauges, rainfall stations, and pump stations to enhance forecasting accuracy. Sensitivity analysis identified ten key parameters influencing model performance, which were adjusted during calibration. The model was calibrated and validated using historical flood events from 2019 to 2022, achieving strong performance metrics, including Nash-Sutcliffe Efficiency (NSE) greater than 0.65, Coefficient of Determination (R²) above 0.80, and Mean Absolute Error (MAE) less than 0.1 meters. The system was further tested using significant flood events from 2020 and 2021, cross-validated with flood marks and community-reported observations, ensuring reliability. This advanced urban flood forecasting tool supports decision-making and emergency response by providing early warnings and flood scenarios, enabling proactive flood management for a rapidly urbanizing city like Dhaka.

While increasing empirical evidence suggests that global time series forecasting models can achieve better forecasting performance than local ones, there is a research void regarding when and why the global models fail to provide a good forecast. This paper uses anomaly detection algorithms and Explainable Artificial Intelligence (XAI) to answer when and why a forecast should not be trusted. To address this issue, a dashboard was built to inform the user regarding (i) the relevance of the features for that particular forecast, (ii) which training samples most likely influenced the forecast outcome, (iii) why the forecast is considered an outlier, and (iv) provide a range of counterfactual examples to understand value changes, in the feature vector or the predicted value, can lead to a different outcome. Moreover, a modular architecture and a methodology were developed to iteratively remove noisy data instances from the train set, to enhance the overall global time series forecasting model performance. Finally, to test the effectiveness of the proposed approach, it was validated on two publicly available real-world datasets.

Time series data are significant and are derived from temporal data, which involve real numbers representing values collected regularly over time. Time series have a great impact on many types of data. However, time series have anomalies. We introduce hybrid algorithm named novel matrix profile (NMP) to solve the all-pairs similarity search problem for time series data. The proposed NMP inherits the features from two state-of-the art algorithms: similarity time-series automatic multivariate prediction (STAMP), and short text online microblogging protocol (STOMP). The proposed algorithm caches the output in an easy-to-access fashion for single- and multidimensional data. The proposed NMP algorithm can be used on large data sets and generates approximate solutions of high quality in a reasonable time. The proposed NMP can also handle several data mining tasks. It is implemented on a Python platform. To determine its effectiveness, it is compared with the state-of-the-art matrix profile algorithms i.e., STAMP and STOMP. The results confirm that the proposed NMP provides higher accuracy than the compared algorithms.

The removal of non-climatic biases, so-called inhomogeneities, from long climatic records needs sophistically developed statistical methods. One principle is that usually the differences between a candidate series and its neighbour series are analysed instead of directly the candidate series, in order to neutralize the possible impacts of regionally common natural climate variation on the detection of inhomogeneities. In most homogenization methods, two main kinds of time series comparisons are applied, i.e. composite reference series or pairwise comparisons. In composite reference series the inhomogeneities of neighbour series are attenuated by averaging the individual series, and the accuracy of homogenization can be improved by the iterative improvement of composite reference series. By contrast, pairwise comparisons have the advantage that coincidental inhomogeneities affecting several station series in a similar way can be identified with higher certainty than with composite reference series. In addition, homogenization with pairwise comparisons tends to facilitate the most accurate regional trend estimations. A new time series comparison method is presented here, which combines the use of pairwise comparisons and composite reference series in a way that their advantages are unified. This time series comparison method is embedded into the ACMANT homogenization method, and tested in large, commonly available monthly temperature test datasets.

Time irreversibility, i.e. the lack of invariance of the statistical properties of a system under time reversal, is a fundamental property of all systems operating out of equilibrium. Time reversal symmetry is associated with important statistical and physical properties and is related to the predictability of the system generating the time series. Over the past fifteen years, various methods to quantify time irreversibility in time series have been proposed, but these can be computationally expensive. Here we propose a new method, based on permutation entropy, which is essentially parameter-free, temporally local, yields straightforward statistical tests, and has fast convergence properties. We apply this method to the study of financial time series, showing that stocks and indices present a rich irreversibility dynamics. We illustrate the comparative methodological advantages of our method with respect to a recently proposed method based on visibility graphs, and discuss the implications of our results for financial data analysis and interpretation.

This research paper investigates methodologies for forecasting the demand for railway tickets, a crucial aspect of optimizing resource allocation, enhancing service quality, and increasing revenue in railway transportation systems. The study explores various time series analysis techniques, including ARIMA, SARIMAX, and neural networks such as LSTM and CNN, to predict ticket demand. Railway ticket sales data spanning several years was utilized to train and validate the models. The performance of each model was evaluated based on key metrics like Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE). Demand is usually defined in Origin-Destination (OD) matrices, which represent movements from the origin station to the destination station for all OD pairs. The proposed CNN model predicts the OD matrix, whereas other models predict demand for each OD pair separately. Findings indicate that the CNN model gives more accurate predictions compared to ARIMA, SARIMAX, and LSTM models. Hence, interdependencies between various OD pairs allow for a more holistic and accurate prediction model.

We define a new seasonal forecasting method based on fuzzy transforms. We use the best interpolating polynomial for extracting the trend of the time series and generate the inverse fuzzy transform on each seasonal subset of the universe of discourse for predicting the value of a an assigned output. Like first example, we use the daily weather dataset of the municipality of Naples (Italy) starting from data collected from 2003 till to 2015 making predictions on the following outputs: mean temperature, max temperature and min temperature, all considered daily. Like second example, we use the daily mean temperature measured at the weather station “Chiavari Caperana” in the Liguria Italian Region. We compare the results with our method, the average seasonal variation, ARIMA and the usual fuzzy transforms concluding that the best results are obtained under our approach in both examples.

Deep learning methods have been widely applied to time series classification tasks. Although deep learning methods perform excellently on these tasks, they require much training data. Labeling time series data for training is very time-consuming and labor-intensive. Active learning can be used to select the most informative data for labeling in time series classification tasks to save human labeling efforts. In this paper, we propose a new active learning (AL) framework for time series data. First, a temporal classifier for pixel-level temporal classification tasks is designed. Next, We propose an effective active learning method to select informative time series samples for labeling, which combines representativeness and uncertainty. For representativeness, We use the K-shape method to cluster time series data. For uncertainty, we construct an auxiliary deep network to evaluate the uncertainty of unlabeled data. The features of the middle hidden layer with rich temporal information of the classifier will be fed into the auxiliary deep network, which is equipped with a self-attention mechanism to utilize the temporal dependencies of time series data fully. Then, we define a new loss function with the aim of improving the deep model’s performance. Finally, the proposed method in this paper was evaluated on two temporal remote-sensing image datasets. The results demonstrate a significant advantage of our method over other approaches to time series data. Code is available at https://github.com/Fighting-Golion/time_series_active_learning

Traditional applications of Interferometric Synthetic Aperture Radar (InSAR) data involved inverting an interferogram stack to determine the average displacement velocity. While this approach has useful applications in continuously deforming regions, much information is lost by simply fitting a line through the time series. Thanks to regular acquisitions across most of the the world by the ESA Sentinel-1 satellite constellation, we are now in a position to explore opportunities for near-real time deformation monitoring. In this paper we present a simple statistical approach for detecting offsets and gradient changes in InSAR time series. Our key assumption is that 5 years of Sentinel-1 data is sufficient to calculate the population standard deviation of the detection variables. Our offset detector identifies statistically significant peaks in the first, second and third difference series. The gradient change detector identifies statistically significant movements in the second derivative series. We exploit the high spatial resolution of Sentinel-1 data and the spatial continuity of geophysical deformation signals to filter out false positive detections that arise due to signal noise. When combined with near-real time processing of InSAR data these detectors, particularly the gradient change, could be used to detect incipient ground deformation associated with geophysical phenomena, for example from landslides or volcanic eruptions.

The study of order patterns of three equally spaced values $x_t,x_{t+d},x_{t+2d}$ in a time series is a powerful tool. The lag $d$ is changed in a wide range so that differences of frequencies of order patterns become autocorrelation functions. Similar to a spectrogram in speech analysis, four ordinal autocorrelation functions are used to visualize big data series, as for instance heart and brain activity over many hours. The method applies to real data without preprocessing, outliers and missing data do not matter. On the theoretical side, we study properties of order correlation functions and show that the four autocorrelation functions are orthogonal in a certain sense. An analysis of variance of a modified permutation entropy can be performed with four variance components associated with the functions.

Retailers must have accurate sales forecasts to operate their businesses efficiently and effectively and to remain competitive in the marketplace. Global forecasting models like RNNs can be a powerful tool for forecasting in retail settings, where multiple time series are often interrelated and influenced by a variety of external factors. By including covariates in a forecasting model, we can better capture the various factors that can influence sales in a retail setting. This can help improve the accuracy of our forecasts and enable better decision-making for inventory management, purchasing, and other operational decisions. In this study we investigate how the accuracy of global forecasting models is affected by the inclusion of different potential demand covariates. To ensure the significance of the study's findings, we used the M5 forecasting competition's openly accessible and well-established dataset. The results obtained from the DeepAR models, which were trained on various combinations of features including time, price, events, and IDs, suggest that individually only the features corresponding to IDs improve the baseline model. However, when all the features are used together, the best performance is achieved, indicating that the individual relevance of each feature is emphasized when the information is given jointly. Comparing the model with features to the model without features, there is an improvement of 1.76\% for MRMSSE and 6.47\% for MMASE.

Fluctuating stock prices make it difficult for investors to see investment opportunities. One tool that can help investors overcome this is forecasting techniques. Long Short-Term Memory (LSTM) is one of deep learning methods used in forecasting time series. The training and success of deep learning is strongly influenced by the selection of hyperparameters. This research uses a hybrid method between the Genetic Algorithm (GA) and LSTM to find a suitable model for predicting stock prices. GA is used in optimizing the architecture such as the number of epochs, window size, and the number of LSTM units in the hidden layer. Tuning optimizer is also carried out using several optimizers to achieve the best value. From method that has been applied, it shows that the method has a good level of accuracy with MAPE values below 10% in every optimizer used. The error rate generated is quite low, in case-1 with a minimum RMSE value of 93.03 and 94.40, & in case-2 with an RMSE value of 104.99 and 150.06 during training and testing. A fairly stable and small value is generated by setting it using the Adam Optimizer.

Machine learning methods have widely been applied to detect anomalies in machine and cutting tool behavior during lathe or milling. However, detecting anomalies in the workpiece itself did not get the same attention by researchers. That is why in this article, the authors present a pub-licly available multivariate time series dataset which was recorded during milling of 16MnCr5. Due to artificially introduced, though realistic anomalies in the workpiece the dataset can be ap-plied for anomaly detection. By using a convolutional autoencoder as a first model good results in detecting the location of the anomalies in the workpiece were achieved. Furthermore, milling tools with two different diameters where used which led to a dataset eligible for transfer learn-ing. The objective of this article is to provide researchers with a real-world time series dataset of the milling process which is suitable for modern machine learning research topics like anomaly detection and transfer learning.

In this paper, we study the asymptotic behavior of the sequential empirical process and the sequential empirical copula process, both constructed from residuals of multivariate stochastic volatility models. Applications for the detection of structural changes and specification tests of the distribution of innovations are discussed. It is also shown that if the stochastic volatility matrices are diagonal, which is the case if the univariate time series are estimated separately instead of being jointly estimated, then the empirical copula process behaves as if the innovations were observed; a remarkable property. As a by-product, one also obtains the asymptotic behavior of rank-based measures of dependence applied to residuals of these time series models.

Hand and arm gesture recognition using radio frequency (RF) sensing modality proves valuable in man-machine interface and smart environment. In this paper, we use time series analysis method for accurately measuring the similarity of the micro-Doppler (MD) signatures between the training and test data, thus providing improved gesture classification. We characterize the MD signatures by the maximum instantaneous Doppler frequencies depicted in the spectrograms. In particular, we apply the dynamic time warping (DTW) method and compare its performance with that of the long short-term memory (LSTM) network. Both methods take into account the values as well as the temporal evolution and trends of time series data. It is shown that the DTW method achieves high gesture classification rates and is robust to time misalignment.

Noise model selection criteria has a significant impact on identifying the stochastic noise proper-ties of any GNSS daily coordinate time series. The low-frequency random walk noise existing in these time series could lead to overestimation of the tectonic rate, so it is of great significance to accurately detect the random walk component. This study focuses on noise model estimation cri-terion (BIC_tp) derived from the AIC and the BIC by introducing 2π factors. It is more sensitive to abnormal steps (random jumps). Using observation data from 72 GNSS stations from 1992 to 2022 and simulated data, four combined noise models are used to explore the impacts of time se-ries lengths (ranging from2 to 24 years) and data loss (between 2% and 30%) on noise models and velocity estimation. The results show that as the time length increased, the selected optimal noise model, and the estimated uncertainty of the tectonic trend with different data gap, gradually con-verge. When the time length is short (less than 8 years), it could lead to the FNRWWN, FNWN, and PLWN models being mistakenly estimated as GGMWN models, thereby affecting the accura-cy of determining the station velocity parameters. When the time length is 12 years, the RW noise component is more probably detected, As the time length increases, the impact of RW on velocity uncertainty is weakened. Finally, we conclude that for a time series with a minimum time length of 12 years, both the selection of the optimal stochastic noise model and the estimation of the ve-locity parameters are reliable.

We propose a novel tensor-based formalism for inferring causal structures from time series. An information theoretical analysis of transfer entropy (TE), shows that TE results from transmission of information over a set of communication channels. Tensors are the mathematical equivalents of these multi-channel causal channels. A multi-channel causal channel is a generalization of a discrete memoryless channel (DMC). We consider a DMC as a single-channel causal channel. Investigation of a system comprising three variables shows that in our formalism, bivariate analysis suffices to differentiate between direct and indirect relations. For this to be true, we have to combine the output of multi-channel causal channels with the output of single-channel causal channels. We can understand this result when we consider the role of noise. Subsequent transmission of information over noisy channels can never result in less noisy transmission overall. This implies that a Data Processing Inequality (DPI) exists for transfer entropy.

This paper applies a new artificial intelligence architecture, the Temporal Fusion Transformer (TFT), for the joint GDP forecasting of 25 OECD countries at different time horizons. This new attention-based architecture offers significant advantages over other deep learning methods. First, results are interpretable since the impact of each explanatory variable on each forecast can be calculated. Second, it allows to visualize persistent temporal patterns and to identify significant events and different regimes. Third, it provides quantile regressions and permits to train the model on multiple time series from different distributions. Results suggest that TFTs outperform regression models, especially in periods of turbulence such as the COVID-19 shock. Interesting economic interpretations are obtained depending on whether the country is domestic demand-led or export-led growth. In essence, TFT is revealed as a new tool that artificial intelligence provides to economists and policy makers, with enormous prospects for the future.

PM-Gati-Shakti Initiative, integration of ministries, including railways, ports, waterways, logistic infrastructure, mass transport, airports, and roads. Aimed at enhancing connectivity and bolstering the competitiveness of Indian businesses, the initiative focuses on six pivotal pillars known as "Connectivity for Productivity": comprehensiveness, prioritization, optimization, synchronization, analytical, and dynamic. In this study, we explore the application of these pillars to address the problem of "Maximum Demand Forecasting in Delhi." Electricity forecasting plays a very significant role in the power grid as it is required to maintain a balance between supply and load demand at all times, to provide a quality electricity supply, for Financial planning, generation reserve, and many more. Forecasting helps not only in Production Planning but also in Scheduling like Import / Export which is very often in India and mostly required by the rural areas and North Eastern Regions of India. As Electrical Forecasting includes many factors which cannot be detected by the models out there, We use Classical Forecasting Techniques to extract the seasonal patterns from the daily data of Maximum Demand for the Union Territory Delhi. This research contributes to the power supply industry by helping to reduce the occurrence of disasters such as blackouts, power cuts, and increased tariffs imposed by regulatory commissions. The forecasting techniques can also help in reducing OD and UD of Power for different regions. We use the Data provided by a department from the Ministry of Power and use different forecast models including Seasonal forecasts for daily data.

The high penetration of distributed energy resources poses significant challenges to the dispatch and operation of power systems. Improving the accuracy of short-term load forecasting (STLF) can optimize grid management, leading to increased economic and social benefits. Currently, some simple AI and hybrid models has issues to deal with and struggle with multivariate dependencies, long-term dependencies, and non-linear relationships. This paper proposes a novel hybrid model for short-term load forecasting (STLF) that integrates multiple AI models with Lasso regression using the stacking technique. The base learners include ANN, XgBoost, LSTM, Stacked LSTM, and Bi-LSTM, while Lasso regression serves as the meta-learner. By considering factors such as temperature, rainfall, and daily electricity prices, the model aims to more accurately reflect real-world conditions and enhance predictive accuracy. Empirical analyses on real-world datasets from Australia and Spain show significant improvements in forecasting accuracy, with a substantial reduction in Mean Absolute Percentage Error (MAPE) compared to existing hybrid models and individual AI models. This research highlights the efficiency of the stacking technique in improving STLF accuracy, suggesting potential operational efficiency benefits for the power industry.

In this study, we present a statistical forecasting framework and assess its efficacy using a range of established machine learning algorithms for predicting Particulate Matter (PM) concentrations in the Arctic, specifically in Pallas (FI), Reykjavik (IS), and Tromso (NO). Our framework leverages historical ground measures and 24-hour predictions from 9 models provided by the Copernicus Atmosphere Monitoring Service (CAMS) to provide PM predictions for the following 24 hours. Furthermore, we compare the performance of various memory cells based on artificial neural networks (ANN), including recurrent neural networks (RNNs), gated recurrent units (GRUs), long short-term memory networks (LSTMs), echo state networks (ESNs), and windowed multi-layer perceptrons (MLPs), which are commonly employed in time series forecasting tasks. Irrespective of the chosen memory cell type, our results demonstrate that the proposed framework consistently outperforms the CAM models in terms of mean squared error (MSE), with average improvements ranging from 25% to 40%. Additionally, we investigate the impact of outliers on the overall model performance.

Industrial data scarcity is one of the largest factors holding back the widespread use of machine learning in manufacturing. To overcome this problem, the concept of transfer learning was developed and it achieved high attention in recent industrial research. Our paper focuses on the problem of time series segmentation and presents the first in-depth research about transfer learning for deep-learning based time series segmentation on the example of industrial end-of-line pump testing. In particularly, we investigate if the performance of deep learning models can be increased by pretraining the network with data from other domains. Three different scenarios are analyzed: source and target data being closely related, source and target data being distantly related, and source and target data being non-related. The results demonstrate that transfer learning can enhance the performance of time series segmentation models in respect to accuracy and training speed. The benefit is most clearly seen in scenarios where source and training data are closely related and the number of target training data samples is lowest. However, in the scenario of non-related datasets, cases of negative Transfer Learning were observed as well. Thus, the research emphasizes the potential, but also the challenges of industrial Transfer Learning.

In this paper we compare the effects of forecasting demand using individual (disaggregated) components versus first aggregating the components either fully or into several clusters. Demand streams are assumed to follow autoregressive moving average (ARMA) processes. Using individual demand streams will always lead to a superior forecast compared to any aggregates, however we show that if several aggregated clusters are formed in a structured manner then these subaggregated clusters will lead to a forecast with minimal increase in mean-squared forecast error. We show this result based on theoretical MSFE obtained directly from the models generating the clusters as well as estimated MSFE obtained directly from simulated demand observations. We suggest a pivot-algorithm, that we call Pivot Clustering, to create these clusters. We also provide theoretical results to investigate sub-aggregation, including for special cases such as, aggregating MA(1) streams and ARMA streams with similar or same parameters.

This paper has three objectives, the first is to present a detailed overview in the form of a tutorial for the developments of several key quantile time series modelling approaches. The second objective is to develop a general framework to represent such quantile models in a unifying manner in order to easily develop extensions and connections between existing models that can then be developed to further extend these models in practice. In this regard, the core theme of the paper is to provide perspectives to a general audience of core components that go into construction of a quantile time series model and then to explore each of these core components in detail. The paper is not addressing the concerns of estimation of these models, as there is existing literature on these aspects in many settings, we provide references to relevant works on these aspects in several classes of model. Instead, the focus is rather to provide a unified framework to construct such models for practitioners, therefore the focus is instead on the properties of the models and links between such models from a constructive perspective. The third objective is to compare and discuss the application of the different quantile time series models on several sets of interesting demographic and mortality based time series data sets of relevance to life insurance analysis. The exploration included detailed mortality, fertility, births and morbidity data in several countries with more detailed analysis of regional data in England, Wales and Scotland.

Building electric energy is characterized by a significant increase of its uses (e.g. vehicle charging), a rapidly declining cost of all related data collection and a proliferation of smart grid concepts, including diverse and flexible electricity pricing schemes. Not surprisingly, an increased number of approaches have been proposed for its modeling and forecasting. In this work, we place our emphasis on three forecasting related issues. First, on the forecasting explainability, i.e. the ability to understand and explain to the user what shapes the forecast. To this extent we rely on concepts and approaches that are inherently explainable, such as the evolutionary approach of genetic programming (GP) and its associated symbolic expressions, as well as the so-called SHAP (SHapley Additive eXplanations) values, which is a well established model agnostic approach for explainability, especially in terms of feature importance. Second, we investigate the impact of the training timeframe on the forecasting accuracy; this is driven by the realization that a fast training would allow for faster deployment of forecasting in real life solutions. And third, we explore the concept of counterfactual analysis on actionable features, i.e. features that the user can really act upon and which therefore present an inherent advantage when it comes to decision support. We have found that SHAP values can provide important insights into the model explainability. As for GP models, we have found comparable and in some cases superior accuracy when compared to its neural-network and time-series counterparts but a rather questionable potential to produce crisp and insightful symbolic expressions, allowing a better insight into the model performance. We have also found, and report here on an important potential especially for practical, decision support, solutions of counterfactuals built on actionable features and short training timeframes.

Consumer-grade economical radon monitors are becoming increasingly popular in private and institutional use, both in the context of Citizen Science and traditional research. Although originally designed for screening indoor radon levels in view of radon regulation and decisions about mitigation or remediation - motivated by the health hazard posed by high radon concentrations - researchers are increasingly exploring their potential in some environmental studies. For long time, radon has been used as a tracer for investigating atmospheric transport processes. This paper focuses on the RadonEye, currently the most sensitive among low-cost monitors available on the market, and specifically, its potential use for monitoring very low radon concentrations. It has two objectives: firstly, discussing issues of statistics of low count rates and secondly, analyzing radon concentration time series acquired with RadonEyes outdoors and in low-radon indoor spaces. Regarding the first objective, among other things, the inference of reported to expected true radon concentration is discussed. The second objective includes the application of autoregressive methods and fractal statistics to time series analysis. The overall result is that radon dynamics can be well captured using this "low-tech" approach. Statistical results are plausible; however, few results are available in the literature for comparison, particularly concerning fractal methods. The paper may therefore be seen as an incentive for further research in this direction.

A continuous Deep Belief Network (cDBN) with two hidden layers is proposed in this paper, focusing on the problem of weak feature learning ability when dealing with continuous data. In cDBN, the input data is trained in an unsupervised way by using continuous version of transfer functions, the contrastive divergence is designed in hidden layer training process to raise convergence speed, an improved dropout strategy is then implemented in unsupervised training to realize features learning by de-cooperating between the units, and then the network is fine-tuned using back propagation algorithm. Besides, hyper-parameters are analysed through stability analysis to assure the network can find the optimal. Finally, the experiments on Lorenz chaos series, CATS benchmark and other real world like CO2 and waste water parameters forecasting show that cDBN has the advantage of higher accuracy, simpler structure and faster convergence speed than other methods.

The purpose of this project is to explore the potential of Long Short-Term Memory (LSTM), a type of recurrent neural network (RNN), as a tool for making directional predictions for assets. The focus will be on using LSTM for classifying up/down moves in stock indices, stocks, and FX rates (EUR/USD). Various financial ratios, advanced technical indicators, and volatility estimators will be used as features to enhance the predictive power of the model. Incorporating credit spreads and news indicators can also improve the model. The advantage of LSTM is its ability to handle sequences of arbitrary length, which means that longer time frames such as 5-day, 10-day, or 1-week returns can be used for prediction. Before using RNNs, exploratory data analysis (EDA) will be performed to understand the underlying patterns in the data. This may include dimensionality reduction techniques such as autoencoding, self-organizing maps, or decision tree regression. The results of the EDA will be used to guide the selection of features for the LSTM model. It is important to note that long memory and stationarity are common characteristics of time series data, such as interest rates or economic indicators. Power-law autocorrection, Hurst exponent fractional, or Markov processes may be used to model these data. Predicting 5-day, 10-day returns for equities or 1-week, 1-month returns for financial factors presents two challenges. First, a large amount of data is required, typically 7–10 years of data. Second, positive autocorrelation in the 5-day or 10-day returns in equity time series must be isolated.

This study explores the shear-induced reversible gel behavior of suspensions containing silica na-noparticles and poly(ethylene oxide) (PEO). When subjected to vigorous shaking, these mixtures exhibit interesting shear thickening, transitioning from low-viscosity fluids to 'shake-gels' and then reverts to liquid after some time. Rheological measurements conducted at constant shear rates re-veal a significant increase in viscosity by several orders of magnitude occurring at specific time points during the process. Followed by the lowering of the shear rate implemented to understand the relaxation behavior. The investigation specifically examines the influence of various ions from the Hofmeister series on the gelation and relaxation dynamics of these suspensions. Results indicate that salts positioned on the right-hand side of the Hofmeister series, known as salting-in agents, facilitate quicker gelation and slower relaxation of the shake-gels. Conversely, salts on the left-hand side, referred to as salting-out agents, lead to slower gelation and faster relaxation. These observa-tions are attributed to the distinct effects that different salts exert on the conformation of PEO chains. Salting-in agents induce swelling of the polymer chains, enhancing their ability to interact and bridge between silica nanoparticles effectively, thus promoting rapid gelation and sustaining the gel structure for extended periods. In contrast, salting-out agents cause the polymer chains to collapse, reducing their bridging efficiency and resulting in delayed gelation along with a more rapid return to the fluid state during relaxation. Understanding the role of specific ions in modu-lating the gelation and relaxation behavior of colloidal-polymer suspensions provides valuable in-sights for tailoring material properties in various applications, including drug delivery, oil drilling, environmental restoration, coatings, and soft matter engineering.

Time series (TS)-based predictions are made by examining the behaviour of historical data to forecast future values. Multiple industries; such as stock market trading, power load forecasting, medical monitoring, and intrusion detection; frequently use it. Several variables; such as the performance of other markets as well as the economic situation of a country which affect its market performance; significantly affect the prediction of stock market prices. Therefore, this present study uses numerous variables; such as the opening, lowest, highest, and closing prices; to predict the indices of the stock market of the Kingdom of Saudi Arabia (KSA). Successfully accomplishing an investment goal largely depends on choosing the right stocks to buy, sell, or maintain. The project's output is the projected closing prices (regression) over the next seven days, which helps investors make the best decisions. This present study used exponential smoothing (ES) to remove noise from the input data obtained from the Saudi Stock Exchange, or Tadawul, before using a multivariate long short-term memory (LSTM) deep learning (DL) algorithm to forecast stock market prices. The proposed multivariate LSTMDL model had satisfactory prediction rates of 97.49% and 92.19% for the univariate model. Therefore, it can be used to effectively predict stock market prices. The results also indicate that DL as well as multiple sources of information can be used to accurately predict stock markets.

This work focuses on analyzing acoustic emission (AE) signals as a means to predict failure in structures. Two main approaches are considered: (i) long-range correlation analysis using both the Hurst (H) and the Detrended Fluctuation Analysis (DFA) exponents, and (ii) natural time domain (NT) analysis. These methodologies are applied to the data collected from two application examples: a glass fiber reinforced polymeric plate and a spaghetti bridge model, where both structures were subjected to increasing loads until collapse. A traditional (AE) signal analysis is also performed to reference the study of the other methods. Results indicate that the proposed methods yield a reliable indication of failure in the studied structures.

Fractal analysis of time series aims to extract the hidden or complex structures issued from the associated real data, such as scaling laws, non-stationary, dynamics, and so on. In the present work, the air quality index and its components are considered as time series owing fractal structures. To understand their underlying dynamics, wavelet multifractal approach is applied starting with a wavelet multifractal spectrum test to confirm the multifractality of the underlying data. Next, a new model is introduced improving the classical minmax one, provided with a statistical measure for the adequacy of the new model. The theoretical findings are applied to the north west Tabuk region of Saudi Arabia for environmental and climate data related to pollutants and factors involved in the estimation of the air quality index.

Landslide is a sliding movement of rock mass, debris and soil along the slope under the action of gravity. Small Baseline Subset (SBAS) is an established method for the investigation and monitoring of landslide moments. This study focuses on monitoring the long-temporal displacement of mountainous terrain in Danba County, Sichuan Province via SBAS technique, based on 31 scenes of L-band ALOS/PALSAR data from Feb. 2007 to Oct. 2010.The results show that the largest velocity rates in LOS direction are ±120 mm/yr and maximum accumulated displacement is up to -300, which indicates fast movement of the mountainous terrain during the observation time. These results get good consistency against the results of previous study. This demonstrates the strong potential of SBAS technique for monitoring the landslides geohazard.

We investigate the power of time series analysis based on a variety of information-theoretic approaches from statistics (AIC, BIC) and machine learning (Minimum Message Length) - and we then compare their efficacy with traditional time series model and with hybrids involving deep learning. More specifically, we develop AIC, BIC and Minimum Message Length (MML) ARMA (autoregressive moving average) time series models - with this Bayesian information-theoretic MML ARMA modelling already being new work. We then study deep learning based algorithms in time series forecasting, using Long Short Term Memory (LSTM), and we then combine this with the ARMA modelling to produce a hybrid ARMA-LSTM prediction. Part of the purpose of the use of LSTM is to seek capture any hidden information in the residuals left from the traditional ARMA model. We show that MML not only outperforms earlier statistical approaches to ARMA modelling, but we further show that the hybrid MML ARMA-LSTM models outperform both ARMA models and LSTM models.

In this work, a novel approach is presented for the multivariate prediction of hydrological variables, employing a specialized type of recurrent neural network (RNN) known as the long-term memory network (LSTM).The central objective is to develop a flood prediction model using data collected from two hydrological stations situated along the Atrato River in the Choc´o Department, Colombia. The interconnected variables of water level, water flow, and water precipitation are analyzed in this model to enhance flood prediction accuracy. The LSTM model, integrated into an RNN structure, is designed to capture the complex dynamics and cross-correlations inherent in these hydrological variables. Validation involves a comparison with linear and nonlinear Nonlinear AutoRegressive with eXogenous input (NARX) models, considering factors such as estimation error and computational time. Additionally, the study conducts further analysis for water level prediction under two scenarios: utilizing only outflow and inflow measurements or predicting outflow alongside measuring inflow, with the objective of utilizing the proposed approach as a flood early warning system. † This article is a revised and expanded version of a paper entitled Multivariable NARX based Neural Networks Models for Short-term Water Level Forecasting, which was presented at International Conference on Time Series and Forecasting (ITISE-2023), Gran Canaria, Spain, 12th-14th July, 2023.

Lake Malombe fish stocks have been depleted by chronic overfishing. Various management approaches (co-management, command control, and ecosystem-based management to fisheries) have been used to manage the fishery. However, the lack of an accurate predictive model has hampered their success. Therefore, we developed and tested a time series model for Lake Malombe fishery. The seasonal fish biomass and CPUE trends were first observed and both were non-stationary. The second-order differencing was applied to transform the non-stationary data into stationary. Autocorrelation functions (AC), partial autocorrelation function (PAC), and Akaike information criterion (AIC) were estimated, which led to the identification and construction of autoregressive integrated moving average (ARIMA) models, suitable in explaining the time series and forecasting. The results showed that ARIMA (1,2,1) provided a better prediction than its counterparts. The model satisfactorily predicted that by 2032, both fish biomass and CPUE will decrease to 3204.6 tons and 59.672 respectively, signifying the potential threat to Lake Malombe fishery. The model justified the necessity of taking precautionary measures to avoid the total collapse of the fishery.

Background: Digital clinical measures collected via various digital sensing technologies such as smartphones, smartwatches, wearables, ingestible, and implantable are increasingly used by individuals and clinicians to capture health outcomes or behavioral and physiological characteristics of individuals. Time series classification (TSC) is very commonly used in modeling digital clinical measures. While deep learning models for TSC are very common and powerful, there exist some fundamental challenges. This review presents non-deep learning models commonly used for time series classification in biomedical applications that achieve high performance. Objective: We performed a systematic review to characterize the techniques used in time series classification of digital clinical measures throughout all stages of data processing and model building. Methods: We conducted a literature search on PubMed, and the Institute of Electrical and Electronics Engineers (IEEE), Web of Science, and SCOPUS databases using a range of search terms to retrieve peer-reviewed articles reporting academic research on digital clinical measures in the five year period between June 2016 and June 2021. We identified and categorized research studies based on the types of classification algorithms and sensor input types. Results: We found 452 papers in total from four different databases: PubMed, IEEE, Web of Science Database, and SCOPUS. After removing duplicates and irrelevant papers, 135 articles remained for detailed review and data extraction. Among these, engineered features using time series methods that were subsequently fed into widely-used machine learning classifiers was the most commonly used technique and also most frequently achieved the best performance metrics (77 out of 135 articles). Statistical modeling (24 out of 135 articles) algorithms were the second most common and also second best classification technique. Wavelet-based classification models (8 out of 135 articles) were also common. Electroencephalogram (29 out of 135 articles) was the most common data type used as an input. Accuracy was the most commonly reported performance metric, with 67.65% of articles reporting on accuracy. In this review paper, we provide summaries of signal pre-processing methods, feature engineering and selection methods, time series models, as well as model interpretations. Importantly, we found that about 50% of the papers only report one performance metric, which may result in a skewed view of overall performance. Conclusion: While high time series classification performance has been achieved in digital clinical, physiological, or biomedical measures, no standard benchmark datasets, modeling methods, or reporting methodology exist. There is no single widely used method for time series model development or feature interpretation– many different methods have proven successful.

Landsat and Sentinel-2 data archives provide ever-increasing amounts of satellite data for studying land cover and land use change (LCLUC) over the past four decades. However, the availability of cloud-, shadow-, and snow-free observations varies spatially and temporally due to climate and satellite data acquisition schemes. Spatio-temporal heterogeneity poses a major issue for some time-series analysis approaches, but can be addressed with pixel-based compositing that generates temporally equidistant cloud-free or near-cloud free synthetic images. Although much consideration is given to methods identifying the ‘best’ pixel value for each composite, determining the aggregation period receives less attention and is often done arbitrary, or based on expert intuition. Here, we evaluated data compositing windows ranging from five days to one year for 1984-2021 Landsat and 2015-2021 Sentinel‑2 time series across Europe. We considered separate and joint use of both data archives and analyzed spatio-temporal availability of composites during each calendar year and pixel-specific growing season. We reported mean annual composites’ availability investigating differences among biogeographical regions, checked feasibility of pan‑European analyses for three LCLUC applications based on annual, monthly and 10-day composites, and analyzed the shortest feasible compositing window ensuring ≥50% temporal data availability and interpolation of the remaining composites for individual years and across a variety of medium- and long‑term time windows. Our results highlighted low data coverage in the 1980s, 1990s, and in 2012, as well as spatial variability in data availability driven by climate and orbit overlaps, which altogether impact spatio-temporal consistency of medium- and long-term time series, limiting feasibility of some LCLUC analyses. We demonstrated that prior to 2011 monthly composites ensured overall 50-62% data coverage for each calendar year, and ~75% afterwards, with further increase to ~82% when Landsat and Sentinel-2 were combined. Temporal consistency of monthly composites was overall low and temporal interpolation augmenting up to 50% missing data each year and across a time window of interest, ensured feasibility of analyses. Applications based on shorter than monthly composites were challenging without joining Landsat and Sentinel‑2 archives after 2015, and beyond the Mediterranean biogeographical region. Using pixel-specific growing season data typically boosted data availability in most geographies and diminished most of the latitudinal differences, but feasibility of complete time series with sub-monthly compositing windows was still restricted to the most recent years, and required data interpolation. Overall, our analyses provided a detailed assessment of Landsat and Sentinel-2 data availability over Europe, and based on selected application examples, highlighted often lacking spatio-temporal consistency of time series with sub-monthly compositing windows and long-time periods, which might hinder feasibility of some LCLUC applications.

The electric power distribution utility is responsible for providing energy to consumers in a continuous and stable way, failures in the electrical power system reduce the reliability indexes of the grid, directly harming its performance. For this reason, there is a need for failure prediction to reestablish power in the shortest possible time. Considering an evaluation of the number of failures over time, this paper proposes to perform a failure prediction during the first year of the pandemic in Brazil (2020) to verify the feasibility of using time series forecasting models for fault prediction. The Long Short-Term Memory (LSTM) model will be evaluated to obtain a forecast result that can be used by the electric power utility to organize the maintenance teams. The Wavelet transform shows to be promising in improving the predictive ability of the LSTM, making the Wavelet LSTM model suitable for the study at hand. The results show that the proposed approach has better results regarding the evaluation of the error in prediction and has robustness when a statistical analysis is performed.

Smart grids require flexible data driven forecasting methods. We propose clustering tools for bottom-up short-term load forecasting. We focus on individual consumption data analysis which plays a major role for energy management and electricity load forecasting. The two first sections are dedicated to the industrial context and a review of individual electrical data analysis. We are interested in hierarchical time-series for bottom-up forecasting. The idea is to disaggregate the signal in such a way that the sum of disaggregated forecasts improves the direct prediction. The 3-steps strategy defines numerous super-consumers by curve clustering, builds a hierarchy of partitions and selects the best one minimizing a forecast criterion. Using a nonparametric model to handle forecasting, and wavelets to define various notions of similarity between load curves, this disaggregation strategy applied to French individual consumers leads to a gain of 16\% in forecast accuracy. We then explore the upscaling capacity of this strategy facing massive data and implement proposals using R, the free software environment for statistical computing. The proposed solutions to make the algorithm scalable combines data storage, parallel computing and double clustering step to define the super-consumers.

Recent studies have shown the potential of Transformer-based neural networks in increasing prediction capacity. However, classical transformers present several problems such as computational time complexity and high memory requirements, which make Long Sequence Time-Series Forecasting (LSTF) challenging. The contribution to the prediction of time series of flood events using deep learning techniques is examined, with a particular focus on evaluating the performance of the Informer model (a particular implementation of transformer architecture), which attempts to address the previous issues. This study explores the predictive capabilities of the Informer model compared to statistical methods, stochastic models and traditional deep neural networks. The accuracy, efficiency as well as the limits of the approaches are demonstrated via numerical benchmarks relating to real river streamflow applications. Using daily flow data from the River Test in England as the main case study, we conduct a rigorous evaluation of the Informer efficacy in capturing the complex temporal dependencies inherent in flood time series. Among other things, the present work extends its analysis to encompass diverse time series datasets from various locations (>100) in the United Kingdom, providing insights into the generalizability of the Informer. The results highlight the significant superiority of the Informer model over established forecasting methods, especially regarding the LSTF problem. Furthermore, it is observed that the structure of time series, as expressed by climacogram, affects the performance of the Informer network.

This article explores the required amount of time series points from a high-speed traffic network to accurately estimate the Hurst exponent. The methodology consists in designing an experiment using estimators that are applied to time series, followed by addressing the minimum amount of points required to obtain accurate estimates of the Hurst exponent in real-time. The methodology addresses the exhaustive analysis of the Hurst exponent considering bias behavior, standard deviation, mean square error, and convergence using fractional Gaussian noise signals with stationary increases. Our results show that the Whittle estimator successfully estimates the Hurst exponent in series with few points. Based on the results obtained, a minimum length for the time series is empirically proposed. Finally, to validate the results, the methodology is applied to real traffic captures in a high-speed network based on the IEEE 802.3ab standard.

GNSS time series prediction plays a significant role in monitoring crustal plate motion, landslide detection, and maintenance of the global coordinate framework. Long Short-Term Memory (LSTM), a deep learning model has been widely applied in the field of high-precision time series prediction especially when combined with Variational Mode Decomposition (VMD) to form the VMD-LSTM hybrid model. To further improve the prediction accuracy of the VMD-LSTM model, this paper proposes a dual variational modal decomposition long short-term memory (DVMD-LSTM) model to effectively handle the noise in GNSS time series prediction. This model extracts fluctuation features from the residual terms obtained after VMD decomposition to reduce the prediction errors associated with residual terms in the VMD-LSTM model. Daily E, N, and U coordinate data recorded at multiple GNSS stations between 2000 and 2022 are used to validate the performance of the proposed DVMD-LSTM model. The experimental results demonstrate that compared to the VMD-LSTM model, the DVMD-LSTM model achieves significant improvements in prediction performance across all measurement stations. The average RMSE is reduced by 9.86%, and the average MAE is reduced by 9.44%. Furthermore, the average accuracy of the optimal noise model for the predicted results is improved by 36.50%, and the average speed accuracy of the predicted results is enhanced by 33.02%. These findings collectively attest to the superior predictive capabilities of the DVMD-LSTM model, thereby enhancing the reliability of the predicted results.

The future of the Colorado River water supply (WS) affects millions of people and the U.S. economy. A recent study suggested a cross-basin correlation between the Colorado River and its neighboring Great Salt Lake (GSL). Following that study, the feasibility of using the previously developed multi-year prediction of the GSL water level to forecast the Colorado River WS was tested. Time-series models were developed to predict the changes in WS out to 10 years. Regressive methods and the GSL water level data were used for the depiction of decadal variability of the Colorado River WS. Various time-series models suggest a decline in the 10-year-averaged WS since 2013 before starting to increase around 2020. Comparison between this WS prediction and the WS projection published in a 2012 government report (derived from climate models) reveals a widened imbalance between supply and demand by 2020. Further research to update similar multi-year prediction of the Colorado River WS is needed. Such information could aid in management decision making in the face of future water shortages.

Purpose: Decomposition analysis of forecasting errors relating to time series generated by a 3PL logistics operator for ten distribution channels operated by the logistics operator. Design / methodology / approach: The studies were focused on the analysis of 10 distribution channels operated by the 3PL logistics operator who used a forecasting tool based on a modified ARIMA algorithm to prepare forecasts. In this paper, R environment was used. The studies focused on the visual analysis of forecasting error series, on the analysis of the basic parameters of the error time series distributions, on the analysis of STL decomposition and statistical tests relating to trend and seasonality. Findings: The forecasting error analysis indicates that there are different patterns and characteristics of errors for individual channels. The statistical test results for various channels display significant differences between forecast groups in some cases. This suggests that the forecasting tool can be more accurate for some than for other channels. Research limitations: Logistic operations are usually based on numerous variables which may influence forecast quality. Moreover, the absence of any information on the forecasting models and input data used may prevent complete understanding of error sources. Value of the paper: The studies described in this paper emphasized valuable conclusions which can be drawn from the analysis of time series forecasting errors in the context of logistic operations. The findings indicated the need for an adapted approach to forecasting for each and every channel, the importance of improving the forecasting tool and the potential to optimize the forecast accuracy by focusing on the trend and seasonality. For this reason, the analysis is an important input into the theory and practice relating to demand forecasting by logistics operators in distribution networks. The studies contribute to the works related to demand forecasting by logistics operators.

Operational weather and also flood forecasting has been performed successfully for decades and is of great socioeconomic importance. Up to now, forecast products focus on atmospheric variables, such as precipitation, air temperature and, in hydrology, on river discharge. Considering the full terrestrial system from groundwater across the land surface into the atmosphere, a number of important hydrologic variables are missing especially with regard to the shallow and deeper subsurface (e.g. groundwater), which are gaining considerable attention in the context of global change. In this study, we propose a terrestrial monitoring/forecasting system using the Terrestrial Systems Modeling Platform (TSMP) that predicts all essential states and fluxes of the terrestrial hydrologic and energy cycles from groundwater into the atmosphere. Closure of the terrestrial cycles provides a physically consistent picture of the terrestrial system in TSMP. TSMP has been implemented over a regional domain over North Rhine-Westphalia and a continental domain over European in a real-time forecast/monitoring workflow. Applying a real-time forecasting/monitoring workflow over both domains, experimental forecasts are being produced with different lead times since the beginning of 2016. Real-time forecast/monitoring products encompass all compartments of the terrestrial system including additional hydrologic variables, such as plant available soil water, groundwater table depth, and groundwater recharge and storage.

Solar flares are characterized by sudden bursts of electromagnetic radiation from the Sun’s surface, and caused by the changes in magnetic field states in solar active regions. Earth and its surrounding space environment can suffer from various negative impacts caused by solar flares ranging from electronic communication disruption to radiation exposure-based health risks to the astronauts. In this paper, we address the solar flare prediction problem from magnetic field parameter-based multivariate time series (MVTS) data using multiple state-of-the-art machine learning classifiers that include MINImally RandOm Convolutional KErnel Transform (MINIROCKET), Support Vector Machine (SVM), Canonical Interval Forest (CIF), Multiple Representations SEQuence Learner (Mr-SEQL), and Long Short-Term Memory (LSTM)-based deep learning model. Our experiment is conducted on the on the Space Weather ANalytics for Solar Flares (SWAN-SF) benchmark data set, which is a partitioned collection of MVTS data of active region magnetic field parameters spanning over 9 years of operation of the Solar Dynamics Observatory (SDO). The MVTS instances of the SWAN-SF dataset are labeled by GOES X-ray flux-based flare class labels, and attributed to extreme class imbalance because of the rarity of the major flaring events (e.g., X and M). As a performance validation metric in this class-imbalanced dataset, we used the true skill statistic (TSS) score. Finally, we demonstrate the advantages of the MVTS learning algorithm MINIROCKET, which outperformed the aforementioned classifiers without the need for essential data preprocessing steps such as normalization, statistical summarization, and class imbalance handling heuristics.

We have implemented quantum modeling mainly based on Bohmian Mechanics to study time series that contain strong coupling between their events. Compared to time series with normal densities, such time series are associated with rare events. Hence, employing Gaussian statistics drastically underestimates the occurrence of their rare events. The central objective of this study is to investigate the effects of rare events in the probability densities of time series from the point of view of quantum measurements. For this purpose, we first model the non-Gaussian behavior of time series using the multifractal random walk (MRW) approach. Then, we examine the role of the key parameter of MRW, λ, which controls the degree of non-Gaussianity, in quantum potentials derived for time series. Our Bohmian quantum analysis shows that the derived potential takes some negative values in high frequencies (its mean values), then substantially increases, and the value drops again for rare events. Thus, rare events can generate a potential barrier in the high-frequency region of the quantum potential, and the effect of such a barrier becomes prominent when the system transverses it. Finally, as an example of applying quantum potential beyond the microscopic world, we compute quantum potentials for the S&P financial market time series to verify the presence of rare events in the non-Gaussian densities and demonstrate deviation from the Gaussian case.

Vegetation status assessment is crucial for agricultural monitoring and management. Vegetation indices derived from high resolution image time series can be used to derive key phenological parameters for annual crops. In this work, we propose a procedure for the estimation of these parameters and their associated uncertainties. The approach uses Bayesian inference through Markov Chain Monte Carlo in order to obtain the full joint posterior distribution of the phenological parameters given the satellite observations. The proposed algorithm is quantitatively validated on synthetic data. Its use on real data is presented together with an application to real-time within season estimation allowing for phenology forecasting.

Changes in sea level exhibit nonlinearity, nonstationarity, and multivariable characteristics, making traditional time series forecasting methods less effective in producing satisfactory results. To enhance the accuracy of the predictions of changes in sea level, this study introduced an improved VMD–EEMD–LSTM hybrid model. This model decomposes satellite altimetry data from near the Dutch coast using VMD, resulting in components of the Intrinsic Mode Function (IMF) with various frequencies and a residual sequence. EEMD further dissects the residual sequence obtained from VMD into second-order components. These IMFs decomposed by VMD and EEMD are utilized as features in the LSTM model for making predictions, culminating in the final forecasted results. The experimental results demonstrated significant improvements in the predictive performance compared with the VMD–LSTM model. The RMSE (root mean square error) decreased by an average of 58.68%, the MAE (mean absolute error) reduced by an average of 59.96%, and the R2 (coefficient of determination) increased by an average of 49.85% compared with the VMD–LSTM model. In comparison with the EEMD–LSTM model, the RMSE decreased by an average of 26.95%, the MAE decreased by an average of 28.00%, and the R2 increased by an average of 6.53%. The VMD–EEMD–LSTM model exhibited significantly improved predictive performance.

Extensive, but remote oil and gas fields of the United States, Canada, and Russia require the construction and operation of extremely long pipelines. Global warming and local heating effects lead to rising soil temperatures and thus a reduction in the sub-grade capacity of the soils; this causes changes in the spatial positions and forms of the pipelines, consequently increasing the number of accidents. Oil operators are compelled to monitor the soil temperature along the routes of the remoted pipelines in order to be able to perform remedial measures in time. They are therefore seeking methods for the analysis of volumetric diagnostic information. To forecast soil temperatures at the different depths we propose compiling a multidimensional dataset, defining descriptive statistics; selecting uncorrelated time series; generating synthetic features; robust scaling temperature series, tuning the additive regression model to forecast soil temperatures.

Most of existing fuzzy forecasting models partition historical training time series into fuzzy time series and build fuzzy-trend logical relationship groups to generate forecasting rules. The determination process of intervals is complex and uncertainty. In this paper, we present a novel fuzzy forecasting model based on high-order fuzzy-fluctuation trends and the fuzzy-fluctuation logical relationships of the training time series. Firstly, we compare each data with the data of its previous day in historical training time series to generate a new fluctuation trend time series(FTTS). Then, fuzzify the FTTS into fuzzy-fluctuation time series(FFTS) according to the up, equal or down range and orientation of the fluctuations. Since the relationship between historical FFTS and the fluctuation trend of future is nonlinear, Particle Swarm Optimization (PSO) algorithm is employed to estimate the required parameters. Finally, use the acquired parameters to forecast the future fluctuations. In order to compare the performance of the proposed model with that of the other models, we apply the proposed method to forecast the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) time series datasets. The experimental results and the comparison results show that the proposed method can be successfully applied in stock market forecasting or such kinds of time series. We also apply the proposed method to forecast Shanghai Stock Exchange Composite Index (SHSECI) to verify its effectiveness and universality.

The Imbalance Cost Pass-Through (ICPT) is a flexible component of the Incentive Regulation (IBR) that empowers power producers to adjust tariffs in response to variable fuel prices, thereby enhancing the economic resilience of electricity generation. However, the cost of electricity generation is significantly influenced by the volatility of commodity prices, including coal, gas and LNG. In Malaysia, the Energy Commission has conducted biannual reviews of fuel and other generation costs. Any cost savings or increases identified during these reviews will be passed on to customers in the form of rebates or surcharges. Due to that reason, this study proposes ICPT price forecasting for the electricity market in Peninsular Malaysia. The study aims to construct an ICPT-related baseline model for the peninsular generation data by employing three forecasting methods. The forecasting performance is analysed by a Mean Absolute Percentage Error (MAPE). In findings, the ARIMA method is one of the most accurate forecasting methods for fuel prices compared to the Moving Average (MA) and LSSVM methods. The observed price differences between the ARIMA and LSSVM models for ICPT are minimal. The ICPT price for July-December 2022 and January-June 2023 is MYR 0.21/ kWh for ARIMA and MYR 0.18/kWh for LSSVM, which are close to the actual TNB’s ICPT tariff. As for the forecasting, the ICPT price is expected to drop for the next announcement. The findings of this study may contribute to a positive impact on the sustainability of the energy sector in Malaysia.

The prediction of stock prices is a challenging task, particularly for retail investors who may lack the resources and expertise to perform sophisticated quantitative trading. This study focuses on enhancing stock price prediction for retail investors by employing advanced machine learning techniques on data from the stock exchange market. We utilize a comprehensive methodology that includes data preprocessing to handle missing values and outliers, feature engineering, cross-validation, and parameter tuning. The techniques applied include Keras Deep Neural Networks (DNN), LightGBM, LSTM, GRU, and linear regression (LR). Our proposed ensemble model, which combines time series and deep learning models, demonstrates superior performance compared to individual models. This integration of methods leads to significant improvements in prediction accuracy, providing a robust solution for retail investors.

(1) Introduction: The unprecedented coronavirus disease 2019 (COVID-19) epidemic has caused millions of infections worldwide and represents a significant challenge facing modern health care systems. This study was conducted to investigate the impact of lockdown measures on regional neonate services, which might be used to predict the long-term effects of medical behavior. (2) Methods: Using hospital information system (HIS) Statistics data from 4 January 2019 to 27 August 2020, an interrupted time-series analysis was employed to compare changes and trends in hospital admissions and disease spectrum before and after the lockdown interventions. Furthermore, this study was designed to evaluate whether the pandemic influenced newborns' healthcare behavior. (3) Results: Overall, 13,540 infants were admitted to the NICU during the pre-COVID period (n = 12082) and COVID-impacted period (n =4558). The patients' age at admission were younger than that of the pre-COVID-19 period (median age 5d vs. 6d after birth, p＜0.001). The overall number of neonate visits consistently decreased from the first days of the lockdown measures (24 January 2020). The disease spectrum for respiratory system, infectious diseases, and gastrointestinal disease indicated no declined immediately after intervention(p =0.079, p =0.113, p =0.060, respectively). There was an immediate decline in the volume of Jaundice-related conditions (p<0.001) after lockdown measures. The percentage of patients who suffer from respiratory system and infectious diseases has decreased (p =0.005 and p =0.002). However, a relatively high percentage of patients admitted to the neonatal intensive care unit (NICU) presented with Jaundice-related conditions (p <0.001). (4) Conclusions: In summary, the COVID-19 pandemic profoundly impacted the regional neonate services. However, it is still unclear what might be the effect of long-term effects from pandemic.

The enhanced multi-objective symbolic discretization for time series (eMODiTS) method employs a flexible discretization scheme using different value cuts for each non-equal time interval, which requires a computational cost for evaluating each objective function. Therefore, surrogate models were implemented to minimize this disadvantage. Nevertheless, each solution found by eMODiTS is a different-sized vector, so the surrogate model must be able to handle data sets under this characteristic. Consequently, this work's contribution lies in analyzing the surrogate models' implementation on the time series discretization, where each potential scheme is a real-number different-sized vector. For this reason, the surrogate model proposed was k-nearest Neighbors for regression with Dynamic Time Warping as a distance measure. Results suggest our proposal finds a suitable approximation to the final eMODiTS solutions with a functions evaluation reduction rate between 15% and 95%. However, according to Pareto front performance measures, the proposal Pareto front is competitive compared to the eMODiTS Pareto front, reaching an average Generational Distance (GD) between 0.0447 and 0.0536. Moreover, the average Hypervolume Ratio (HVR) ranges between 0.334 and 0.3891. Finally, our proposal compared against SAX-based methods presents a similar behavior regarding classification tasks and statistical tests.

This study investigates the impact of Foreign Direct Investment (FDI) on economic growth and examines the causality between FDI and economic growth in Bangladesh during 1972-2013. Gross Domestic Product (GDP), export performance (EXP), Foreign Direct Investment (FDI), and Gross Fixed Capital Formation (GFCF) are considered to capture the objective of the study. The study methodology includes some systematic steps. As the data used in the study is time-series in nature, the author employs unit root tests, and in this case, Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests are used. Then Johansen’s cointegration test, Granger causality test, regression with Newey-West Standard Error and Vector Error Correction Model (VECM) are applied. By using the ADF and PP test the study reveals that the variables of four-time series are integrated of I (1) i.e. they are stationary at first difference. Regression analysis result demonstrates that FDI has a positive effect on economic growth. The Granger Causality test discloses that there is a unidirectional relationship between FDI and economic growth. But the VECM estimation finds that in the long run FDI negatively affects economic growth.

In Portugal, cork oak (Quercus suber L.) stands cover 737 Mha, being the most predominant species of the montado agroforestry system, contributing for the economic, social and environmental development of the country. Cork oak decline is a known problem since the late years of the 19th century that has recently worsen. The causes of oak decline seem to be a result of slow and cumulative processes, although the role of each environmental factor is not yet established. The availability of Sentinel-2 high spatial and temporal resolution dense time series enables gradual processes monitoring. These processes can be monitored using spectral vegetation indices (VI) once their temporal dynamics are expected to be related with green biomass and photosynthetic efficiency. The Normalized Difference Vegetation Index (NDVI) is sensitive to structural canopy changes, however it tends to saturate at moderate-to-dense canopies. Modified VI have been proposed to incorporate the reflectance in the red-edge spectral region, which is highly sensitive to chlorophyll content while largely unaffected by structural properties. In this research, in-situ data on the location and vitality status of cork oak trees are used to assess the correlation between chlorophyll indices (CI) and NDVI time series trends and cork oak vitality at the tree level. Preliminary results seem to be promising since differences between healthy and unhealthy (diseased/dead) trees were observed.

Stock price prediction has long been a critical area of research in financial modeling. The inherent complexity of financial markets, characterized by both short-term fluctuations and long-term trends, poses significant challenges in accurately capturing underlying patterns. While Long Short-Term Memory (LSTM) networks have shown strong performance in short-term stock price prediction, they struggle with effectively modeling long-term dependencies. In this paper, we propose an advanced stock price prediction model based on the Extended Long Short-Term Memory (xLSTM) algorithm, designed to enhance predictive accuracy over both short and long-term periods. We conduct extensive experiments by building and evaluating models based on xLSTM and LSTM architectures for multiple stocks. Our results demonstrate that the xLSTM model consistently outperforms the LSTM model across all stocks and time horizons, with the performance gap widening as the prediction period extends. The observations underscore the superior capability of the xLSTM-based model to capture long-term patterns in financial data, offering a promising approach for more accurate and reliable stock price predictions.

Climate change and development have increased urban flooding, requiring modernization of stormwater infrastructure. Retrofitting standard passive systems with controllable valves/pumps is promising, but requires real-time control (RTC). One method of automating RTC is reinforcement learning (RL), a general technique for sequential optimization and control in uncertain environments. The notion is that an RL algorithm can use inputs of real-time flood data and rainfall forecasts to learn a policy for controlling the stormwater infrastructure to minimize measures of flooding. In real-world conditions, rainfall forecasts and other state information, are subject to noise and uncertainty. To account for these characteristics of the problem data, we implemented Deep Deterministic Policy Gradient (DDPG), an RL algorithm that is distinguished by its capability to handle noise in the input data. DDPG implementations were trained and tested against a passive flood control policy. Three primary cases were studied: (i) perfect data, (ii) imperfect rainfall forecasts, and (iii) imperfect water level and forecast data. Rainfall episodes (100) that caused flooding in the passive system were selected from 10 years of observations in Norfolk, Virginia, USA; 85 randomly selected episodes were used for training and the remaining 15 unseen episodes served as test cases. Compared to the passive system, all RL implementations reduced flooding volume by 70.5% on average, and performed within a range of 5%. This suggests that DDPG is robust to noisy input data, which is essential knowledge to advance the real-world applicability of RL for stormwater RTC.

This study introduces a novel method that combines Generative Adversarial Networks (GANs) with conformal prediction to revolutionize the domain of stock ticker forecasting. Our approach enhances prediction accuracy and offers dependable prediction intervals to gauge uncertainty, surpassing current GAN-based techniques in unpredictable market scenarios. We made a validation using AAPL stock data, showing the ability of our model to generate accurate forecasts and reliable assessments of prediction uncertainty. The prediction intervals generated provide investors and risk managers with a powerful tool to make informed decisions. Our approach represents a notable advancement in financial prediction as it improves the clarity and reliability of forecasts. By incorporating conformal prediction into the GAN framework, the reliability of forecasts is improved, providing a thorough evaluation of uncertainty that is crucial for successful risk management. This framework has the potential to transform the field of predictive analytics in finance. It shows a solution that balances accuracy and uncertainty quantification. The results clearly show the effectiveness of this method, paving the way for its potential application in a wide range of financial prediction scenarios.

Accurate forecasting and efficient management of commodities are crucial for stakeholders in the industry, given the volatile nature of these markets. This study addresses these needs by leveraging advanced forecasting techniques and machine learning models to predict prices and enhance supply chain efficiency. TThe focus is on utilizing ARIMA, SARIMAX, and LSTM models to analyze historical trading data for commodities such as cocoa, coffee, and sugar sourced from Kaggle’s comprehensive dataset. The research applies ARIMA and SARIMAX models to forecast price trends, overcoming initial challenges related to data index frequency and seasonality. LSTM models are employed for more nuanced demand forecasting, particularly for random-length lumber, demonstrating the model's capability to predict future market trends accurately. The study highlights significant improvements in prediction accuracy and supply chain management through meticulous feature engineering and model optimization, offering valuable insights for strategic decision-making in the supply chain sector.

We present and benchmark FilterNet, a flexible deep learning architecture for time series classification tasks, such as activity recognition via multichannel sensor data. It adapts popular CNN and CNN-LSTM motifs which have excelled in activity recognition benchmarks, implementing them in a many-to-many architecture to markedly improve frame-by-frame accuracy, event segmentation accuracy, model size, and computational efficiency. We propose several model variants, evaluate them alongside other published models using the Opportunity benchmark dataset, demonstrate the effect of model ensembling and of altering key parameters, and quantify the quality of the models’ segmentation of discrete events. We also offer recommendations for use and suggest potential model extensions. FilterNet advances the state of the art in all measured accuracy and speed metrics on the benchmarked dataset, and it can be extensively customized for other applications.

With the gradual development and integration of artificial intelligence into various industries, there is also a great range of integration in the financial industry. Therefore, this article focuses on the trend prediction model and financial risk management problems of deep reinforcement learning (DRL), one of the largest branches of artificial intelligence, in the cryptocurrency market. In addition, in the experimental part of this paper, the artificial intelligence machine learning Long short-term memory network (LSTM) model is used to make effective time series prediction and analysis on the relevant data of the cryptocurrency market, so as to make a large-scale analysis to improve the accuracy of market trend prediction and the effectiveness of risk management. In addition, in this experiment, technology-related indicators, emotional states of financial market customers and other content related to large language models are combined. While optimizing investment strategy by using deep reinforcement learning algorithm, machine learning prediction model is also used to capture the time dependence of financial market. The experimental results also show that the predicted results are consistent with the actual value. Therefore, the model has high practical application value in predicting the time series price trend of cryptocurrency in the financial market and indicates that the integrated DRL model framework can further optimize and manage the price and trading strategy of the financial market. Future research should focus on improving the LSTM model and incorporating more features to improve prediction accuracy and adapt to market changes.

Air sampling using vortex air samplers combined with species specific amplification of pathogen DNA, was carried out over two years in four or five locations in the Salinas Valley of California. The resulting time series data for the abundance of pathogen DNA trapped per day displayed complex dynamics with features of both deterministic (chaotic) and stochastic dynamics. Methods of nonlinear time series analysis developed for the reconstruction of low dimensional attractors provided new insights into the complexity of the pathogen abundance data, but also indicated that practicality may limit the capacity for definitively classifying the dynamics of air borne plant pathogen inoculum. Over the two years of the study five location/year combinations were classified as having stochastic linear dynamics and four were not. Calculation of entropy values for either the number of pathogen DNA copies or for a binary string indicating the pathogen abundance data were increasing or not, revealed (1) some robust differences in the dynamics between seasons that were not obvious in the time series data themselves, and also (2) that the series were almost all at their theoretical maximum entropy value when considered from the simple perspective of whether instantaneous change along the sequence is positive or not.

Coastal wave modeling and forecasting are essential in oceanography, sustainable marine energy, and ocean engineering. Precise forecasting of wave speed and direction are crucial for offshore operations, marine energy, risk management, environmental management, coastal and sustainable maritime management. The coastline of Brunei Darussalam can generate between 15 and 126 Giga Watt of wave energy. In this experimental research, we used two numerical approaches, the finite difference, and spectral element methods, to model and simulate wave speed and direction. We calculated the mean error between numerical and analytical solutions. We proposed a novel, promising, univariate time series forecasting model by combining the Long Short-Term (LSTM) with KerasTuner hyperparameters tuning and optimization techniques. This method helps us to improve the accuracy and efficiency of time-series forecasting.. The experimental data was computed from high-precision Acoustic Doppler Current Profiler (ADCP) sensor data. This research is part of the preliminary feasibility analysis of wave energy production in Brunei Darussalam and net zero commitment for a sustainable environment. Seven independent forecast experiments were performed for wave speed and direction in degree and radian units for 1, 3, 6, 8, 10, 12, and 24 hours. Mean squared error (MSE) was adopted as a metric for both training and testing. The experimental results reveal that the wave speed forecast has the lowest MSE compared to direction, regardless of the unit of measure, but has a longer duration. In addition, the direction forecast in the degree unit has the lowest errors compared to the unit of radians; the latter has a longer running time than the former. The model has delivered optimal results throughout the experiments with minor training and test errors. We conducted a thorough evaluations on two benchmark time series datasets, which include the study dataset and air quality index dataset, to validate the performance of the proposed model with other models. The proposed model outperforms cutting-edge forecasting models, such as the conventional LSTM, ARIMA, and Prophet. The model has the slightest forecast error compared to the existing literature’s result.

Remotely sensed data play a crucial role in monitoring the El-Niño/La-Niña Southern Oscillation (ENSO), which is an oceanic-atmospheric phenomenon occurring quasi-periodically with several impacts worldwide, such as specific biological and global climate responses. Since 1980, Earth has witnessed three strong ENSO events (1982–1983, This study aims, 2015-2016). In September 2022, La-Niña entered its third year and was unlikely to continue through 2023. Instead, since 2022 forecasts pointed to a transition from La-Nina to a Neutral phase in summer or late 2023. The onset of the El-Niño occurred around April 2023, and it is anticipated by sophisticated models to be a strong event through the Northern Hemisphere winter (December 2023 - February 2024). The aim of this study is to demonstrate the ability of the combination of two new methods to improve the accuracy of the above claim because El-Niño apart from climate anomalies, significantly impacts Earth’s ecosystems and human societies, regulating the spread of diseases by insects (e.g., malaria and dengue fever), and influencing nutrients, phytoplankton biomass, and primary productivity. This is done by exploring first the previous major El-Niño events in the period January 1876-July 2023. Our calculations show that the ongoing 2023-2024 El-Niño will not be the strongest.

The accurate forecasting of solar power output is essential for ensuring the stable and efficient operation of grid-connected photovoltaic (PV) systems, especially when faced with variable weather patterns. This research focuses on the integration of Machine Learning (ML) methodologies and climatic parameters to predict solar panel energy generation, with a specific emphasis on addressing consumption-production imbalances. Leveraging a dataset sourced from the Kaggle platform, the study is conducted in the context of Estonia, aiming to optimize solar energy utilization in this geographic region. The dataset, obtained from Kaggle, encompasses comprehensive information on climatic variables, including sunlight intensity, temperature, and humidity, alongside corresponding solar panel energy output. Leveraging machine learning algorithms, including XGBoost regression and neural networks, our model aims to identify complex patterns and relationships within the datasets. By tailoring the model to Estonia's climatic nuances, we seek to enhance the accuracy of energy production forecasts and, consequently, better manage the challenges associated with consumption-production imbalances. Furthermore, the research investigates the adaptability of the proposed model to diverse climatic conditions, ensuring its applicability for similar endeavors in other geographical locations. The performance of the forecasting model will be evaluated using error metrics like mean absolute error (MAE) and root mean squared error (RMSE). This study leverages a comprehensive dataset from Kaggle and utilizes advanced machine learning techniques to generate valuable insights. These insights can inform the development of sustainable energy policies and practices, ultimately leading to a more efficient and reliable renewable energy infrastructure.

In order to reduce the environmental impact of energy enterprises, it is important to promptly identify cases of possible changes in the quality of wastewater generated at power plants, that is cases of exceeding the maximum permissible concentrations of pollutants in wastewater. The goal of the method for detecting complex anomalies in multidimensional time series obtained from smart energy station’s sensor channels is to increase the accuracy of detecting pollution levels in industrial wastewater. To achieve this goal, the following tasks were solved: methods for detecting time series anomalies were analyzed, the method for detecting complex anomalies was developed, software implementation of the algorithm was carried out and experiments were conducted. The developed method is proposed to be used in an smart energy monitoring system.

Time-series analysis is a widely used technique across various fields and industries, as it helps in understanding, predicting, and forecasting the behavior of data points over time. These fields include but are not limited to finance, economics, healthcare, transportation, etc. In the case of this paper, we have focused on finance. Predicting future values of financial time series offers several benefits. Accurate forecasts can help investors make better decisions about their investments. To predict future values, deep learning algorithms are commonly used since it is an effective method with complex data. In this study, we conduct a study that investigates the use of different data pre-processing techniques on deep learning algorithms in predicting the time series values. To conduct this experimental study, we utilize an open source software, which using long short-term memory technique as the representative deep learning technique, published in github software code repository platform. With this study, we investigate the effects of autoencoder and discrete wavelet transform data pre-processing techniques in time-series prediction. We discuss the details of the experimental study and report our results. The results show that time series prediction (using backtesting methodology) without any data pre-processing leads to 12.6% for mean absolute percentage error. The results also show that, time series prediction with the data preprocessing techniques (Wavelet Transform and Stacked Autoencoder) lead to 3.4%

Background: Coronavirus disease 2019 (COVID-19) pandemic are causing significant damages to many nations. For mitigating its risk, Japan’s Prime Minister called on all elementary, junior high and high schools nationwide to close beginning March 1, 2020. However, its effectiveness in decreasing disease burden has not been investigated. Methods: We used daily data on the report of COVID-19 and coronavirus infection incidence in Japan until March 31, 2020. Time series analysis were conducted using Bayesian method. Local linear trend models with interventional effect were constructed for number of newly reported cases of COVID-19, including asymptomatic infections. We considered that the effects of intervention start to appear 9 days after the school closure; i.e., on March 9. Results: The intervention of school closure did not appear to decrease the incidence of coronavirus infection. If the effectiveness of school closure began on March 9, mean coefficient α for effectiveness of the measure was calculated to be 0.08 (95% credible interval -0.36 to 0.65), and the actual reported cases were more than predicted, yet with rather wide credible interval. Sensitivity analyses using different dates also showed similar results. Conclusions: School closure carried out in Japan did not show the effectiveness to mitigate the transmission of novel coronavirus infection.

This paper realizes early bearing fault warning through bearing fault time series prediction, and proposes a bearing fault time series prediction model based on optimized maximum correlation kurtosis deconvolution (MCKD) and long short-term memory (LSTM) recurrent neural network to ensure bearings operation reliability. The model is based on lifecycle vibration signal of the bearing, to begin, the cuckoo search (CS) is utilized to optimize the parameter filter length L and deconvolution period T of MCKD, taking into account the influence and periodicity of the bearing time series, the fault impact component of the optimized MCKD deconvolution time series is improved. Then select the LSTM learning rate α depending on deconvolution time series. Finally, the dataset obtained through various preprocessing approaches are used to train and predict the LSTM model. The average prediction accuracy of the optimized MCKD-LSTM model is 26 percent higher than that of the original time series, proving the efficiency of this method, and the prediction results track the real fault data well, according to the XI'AN JIAOTONG University XJTU-SY bearing dataset.

Land Surface Temperature (LST) is a basic parameter in energy exchange between the land and atmosphere and is frequently used in many sciences such as climatology, hydrology, agriculture, ecology, etc. LST time series data have usually deficient, missing and unacceptable data caused by the presence of clouds in images, presence of dust in atmosphere and sensor failure. In this study, Singular Spectrum Analysis (SSA) algorithm was used to resolve the problem of missing and outlier data caused by cloud cover. The region studied in the present research included an image frame of MODIS with horizontal number 22 and vertical number 05 (h22v05). This image involved a large part of Iran and Turkmenistan and Caspian Sea. In this study, MODIS LST sensor (MOD11A1) was used during 2015 with 1×1 Km spatial resolution and day/night LST data (daily temporal resolution). The results of the data quality showed that cloud cover caused 36.37% of missing data in the studied time series with 730 day/night LST images. Further, the results of SSA algorithm in reconstruction of LST images indicated the Root Mean Square Error (RMSE) of 2.95 K between the original and reconstructed data in LST time series in the study region. In general, the findings showed that SSA algorithm using spatio-temporal interpolation in LST time series can be effectively used to resolve the problem of missing data caused by cloud cover.

The aerospace industry develops prognosis and health management algorithms to ensure better safety on board. Particularly for in-flight controls where jamming is dreaded. For that, vibration signals are monitored to predict future defect occurrences. However, time series are not labeled according to severity level, and the user can only assess the system health from the data mining procedure. To that extent, we developed a clustering algorithm using a deep neural network core. We encoded the time series into pictures to be feed into an artificially trained neural network: U-NET. From the segmented output, one-dimensional information on cluster frontiers is extracted and filtered without any parameter selection. Then, a kernel density estimation finally transforms the signal into an empirical density. Ultimately, a Gaussian mixture model extracts the latter independent components. The method empowered us to reveal different degrees of severity faults in the studied data, with their respective likelihood without prior knowledge. We compared it to state-of-the-art machine learning algorithms. However, internal clustering results evaluation for time series is an open question. As state-of-the-art indexes were not producing relevant results, we built a new indicator to fulfill this task. We applied the whole method to an actuator consisting of an induction machine linked to a ball screw. This study lays the groundwork for future training of diagnosis and prognosis structures in the health management framework.

Hybrid forecasting methods have emerged as a solution surpassing the limitations of both statistical and deep learning approaches. While the first emphasize the significance of variables, they often produce worse forecasting results when compared to newer techniques. In contrast, deep learning models remain enigmatic "black boxes" in terms of interpretability, although achieving better results in forecasting. This article introduces the Comprehensive Cross-Correlation and Lagged Linear Regression Deep Learning (CCLR-DL) framework, designed to harness the best of both approaches, enhancing forecasting accuracy while retaining model interpretability through a feature selection process. CCLR-DL blens cross-correlation analysis, lagged multiple linear regression and granger's causality procedures with deep learning architectures based on LSTM. In a practical demonstration, CCLR-DL was applied to a real database of clinical visits associated to diagnoses in Catalonia, Spain (tracking a population of 6.3 million patients during 10 years). Predicting visits enables the healthcare managers to be ready for future demand shifts. Results demonstrate a consistent and substantial improvement over standalone statistical and deep learning methods when predicting healthcare demand. This hybrid approach not only showcases its efficacy but also offers a promising solution to the challenge of balancing predictive accuracy with model explicability. In this context, this work aims to design and validate a method for feature selection and forecasting of multivariate high dimensional time series datasets not only to improve prediction accuracy but also to model transparency by identifying a subset of variables that improve predictions and G-cause the target variable.

Research within the field of hydrology often focuses on comparing stochastic to machine learning (ML) forecasting methods. The comparisons performed are all based on case studies, while an extensive study aiming to provide generalized results on the subject is missing. Herein, we compare 11 stochastic and 9 ML methods regarding their multi-step ahead forecasting properties by conducting 12 large-scale computational experiments based on simulations. Each of these experiments uses 2 000 time series generated by linear stationary stochastic processes. We conduct each simulation experiment twice; the first time using time series of 100 values and the second time using time series of 300 values. Additionally, we conduct a real-world experiment using 405 mean annual river discharge time series of 100 values. We quantify the performance of the methods using 18 metrics. The results indicate that stochastic and ML methods perform equally well.

In an epoch characterized by the swift pace of digitalization and urbanization, the essence of community well-being hinges on the efficacy of urban management. As cities burgeon and transform, the need for astute strategies to navigate the complexities of urban life becomes in-creasingly paramount. This study employs time series analysis to scrutinize citizen interactions with the coordinate-based problem mapping platform in the Municipality of Patras in Greece. The research explores the temporal dynamics of reported urban issues, with a specific focus on identifying recurring patterns through the lens of seasonality. The analysis, employing the sea-sonal decomposition technique, dissects time series data to expose trends in reported issues and areas of the city that might be obscured in raw big data. It accentuates a distinct seasonal pattern, with concentrations peaking during the summer months. The study extends its approach to forecasting, providing insights into the anticipated evolution of urban issues over time. Projec-tions for the coming years show a consistent upward trend in both overall city issues and those reported in specific areas, with distinct seasonal variations. This comprehensive exploration of time series analysis and seasonality provides valuable insights for city stakeholders, enabling informed decision-making and predictions regarding future urban challenges.

The global demand for energy has been steadily increasing due to population growth, urbanization, and industrialization. Numerous researchers worldwide are striving to create precise forecasting models for predicting energy consumption to manage supply and demand effectively. In this research, a time-series forecasting model based on multivariate multilayered long short-term memory (LSTM) is proposed for forecasting energy consumption and tested using data obtained from commercial buildings in Melbourne, Australia: the Advanced Technologies Center, Advanced Manufacturing and Design Center, and Knox Innovation, Opportunity, and Sustainability Center buildings. This research specifically identifies the best forecasting method for subtropical conditions and evaluates its performance by comparing it with the most used methods at present, including LSTM, bidirectional LSTM, and linear regression. The proposed multivariate multilayered LSTM model was assessed by comparing mean average error (MAE), root-mean-square error (RMSE), and mean absolute percentage error (MAPE) values with and without labeled time. Results indicate that the proposed model exhibits optimal performance with improved precision and accuracy. Specifically, the proposed LSTM model achieved a decrease in MAE by 30%, RMSE by 25%, and MAPE by 20% compared to the LSTM method. Moreover, it outperformed the bidirectional LSTM method with a reduction in MAE by 10%, RMSE by 20%, and MAPE by 18%. Furthermore, the proposed model surpassed linear regression with a decrease in MAE by 2%, RMSE by 7%, and MAPE by 10%. These findings highlight the significant performance increase achieved by the proposed multivariate multilayered LSTM model in energy consumption forecasting.