Predictive Analysis and Optimization in Sustainable Agriculture Facing Climate Change with Emerging Technological Approaches

1. Introduction

Climate change is significantly transforming the way we produce food, creating challenges that directly affect food security worldwide [1]. Changing rainfall patterns, higher temperatures and more frequent extreme events are reducing agricultural yields and making access to food more difficult, especially in the most vulnerable communities [1,2]. These changes not only affect the quantity of what is produced, but also the quality of crops, increasing production costs and limiting their availability in key markets, especially in regions already facing economic and social challenges [3].

For small farmers, who depend on their crops for their livelihoods, these changes are devastating; with limited resources and few options to adapt, they face risks that endanger not only their livelihoods, but also the well-being of their families and communities [4,5]. In many cases, the lack of access to adequate technologies and supporting policies increases their vulnerability to the attacks of climate change [6]. This makes it urgent to find practical and sustainable solutions that not only respond to current challenges, but also allow anticipating future impacts [5].

In the face of these challenges, the optimization of agricultural processes has become an essential strategy to mitigate the effects of climate change [7,8]. This optimization not only seeks to improve the efficiency of agricultural practices, but also to ensure that they are sustainable in the long term [9]. In this context, predictive analysis has emerged as a powerful tool; through the processing of large volumes of data, it is possible to more accurately predict agricultural yields and propose specific interventions that optimize the use of critical resources, such as water, fertilizers and energy [5,8]. For example, the use of IoT (Internet of Things) sensors that collect real-time data on environmental and agricultural variables has allowed farmers to make more informed decisions, thus maximizing productivity and reducing losses [10,11,12].

This article reviews how combining predictive models, such as logistic regression, with mathematical optimization models can revolutionize agriculture in the face of climate change [13,14]. Predictive models help understand how factors such as climate, soil characteristics, and management practices affect crops, while mathematical models offer strategies to maximize results with available resources [15,16]. This approach, which unites science and technology, not only seeks to improve agricultural yields, but also to identify solutions that adapt to the specific needs of each context, considering both the limitations and opportunities of each region [12].

Recent studies show that these tools can make a big difference [14]. For example, research has identified how certain climatic conditions directly impact crops, allowing the design of strategies that reduce losses and increase the capacity to adapt to extreme events [17,18]. These solutions are particularly valuable in regions with high climate vulnerability, where data-driven decisions can mean the difference between the success and failure of a crop [19,20].

The objective of this paper is; first, to critically analyze recent advances in predictive models applied to assessing the impact of climate change on sustainable agriculture. Second, to review optimization techniques, statistical models, and emerging technologies, such as artificial intelligence, IoT sensors, and big data, that are being used to improve the resilience of agricultural systems to climate change [12,21,22].

Ensuring food for all in a context of climate change is not just a technical challenge; it is a commitment to our present and our future [21]. By integrating advanced methodologies such as those mentioned, we not only help farmers face the effects of climate change, but also create stronger, more resilient and prepared agricultural systems for an increasingly uncertain environment [16,23]. Finally, these strategies not only promote global food security, but also strengthen the connection between technology, science and communities that depend on agriculture for their survival [24].

2. methodology

2.1. Type of Study

A systematic review literature study was conducted to analyze optimization and predictive analysis approaches in food production in the context of climate change. The review focused on identifying statistical and mathematical modeling techniques used in recent literature (2013-2024) to maximize agricultural production and manage resources efficiently. This analysis included a critical evaluation of the methods applied, as well as the results obtained in different studies, with the aim of providing a comprehensive and global view of current trends in the field.

2.2. Techniques and Tools

The technique of documentary observation and analysis was used to help us to properly systematize the articles selected with respect to the study. To this end, several indicators were used to conduct a comprehensive assessment of the reviewed literature. The indicators include forecasting techniques used, such as logistic regression and process-based models, which are necessary to forecast agricultural production outcomes. In addition, established optimization methods, from mathematical models to simulations, were used to maximize efficiency in the use of resources. The variables analyzed, including climatic factors and agricultural resources, were also taken into account, as these variables are very important in relation to climate change. A structured observation sheet was used to systematically record relevant data, which helped to gather information on the type of study, the sample, the methods used, and the results obtained, providing a solid basis for analysis and interpretation.

2.3. Literature Search Procedure

The search was conducted in various academic databases, such as Scopus, Web of Science, and Google Scholar. Key terms considered essential for the research were used, such as agricultural optimization, predictive analysis, climate change, statistical models, and logistic regression. To ensure that the process was rigorous and systematic, PRISMA guidelines were followed, adapted to the specific context of this study as shown in Figure 1.

In the first stage, a total of 24 potential articles were identified that were related to the topic of interest. Then, in the selection phase, the abstracts and methodologies of each of these studies were carefully reviewed. In this process, those that did not apply optimization or predictive analysis techniques were excluded, allowing attention to be focused on the most relevant works.

In the next stage, articles that explicitly addressed the combination of statistical and mathematical models to improve agricultural production, especially under adverse climatic conditions, were evaluated. This phase was crucial, as research was identified that really added value to the topic.

Finally, in the inclusion stage, a total of 15 relevant studies that were published between 2013 and 2024 were systematized. This process not only allowed for the gathering of valuable information, but also provided a clearer vision of how optimization and predictive analysis can contribute to facing the challenges of climate change in agriculture. Undoubtedly, each step of this search became a significant learning experience on the academic path.

2.4. Analysis of Studies

The selected articles were analyzed both qualitatively and quantitatively, with the aim of identifying patterns in the use of predictive and optimization methodologies. To facilitate understanding of the findings, the results were organized in tables and graphs that clearly and visually describe the most relevant aspects of each study.

Among the results, the techniques employed, such as the integration of IoT sensors and machine learning, were highlighted, which were mentioned in several articles [1,11]. These emerging technologies are revolutionising the way data is collected and analysed in the field, allowing farmers to make more informed and accurate decisions.

In addition, the regression and climate simulation models used to predict agricultural yields were explored [3,7]. These models are fundamental, as they allow anticipating how variations in the climate can affect production, which is crucial in a context of climate change. Finally, optimization approaches to the allocation of agricultural resources, including water and fertilizers, were examined [2,5,8]. Efficient management of these resources is vital to maximize production and minimize environmental impact, and the studies reviewed offered valuable insights into how to achieve this.

3. Results

Figure 2 presents the distribution of studies on predictive models in sustainable agriculture, classifying them according to the type of model used. The majority of the literature (n=5) focuses on predictive models, followed by studies combining historical and forward-looking perspectives (n=3) and studies focusing on revision (n=4). Only a few studies (n=3) focused on examining historical patterns. The importance of predictive models in the literature reflects the growing interest in using data analytics to improve agricultural sustainability and in using predictions to optimize production and resource management.

Figure 3 shows the use of emerging technologies, optimization techniques, and statistical models in agricultural resilience. There is a predominant use of emerging technologies such as IoT (Internet of Things), AI (Artificial Intelligence), Big Data, and Blockchain, reflecting the increasing integration of digital tools in the agricultural sector to improve management and efficiency. Optimization techniques represented by robust optimization, machine learning algorithm optimization, data-driven optimization, and other methods have been widely used, indicating that people are looking for strategies to achieve maximum efficiency and performance in various agricultural scenarios. Finally, statistical models such as regression, decision trees, and maximum entropy models demonstrate the importance of data-driven predictions and decisions in risk management and adaptation to changes in climate and production conditions.

The reviewed studies from 2013 to 2024 are described in Table 1. A total of fifteen investigations address diverse applications of advanced technologies in agriculture, focusing on yield prediction, crop optimization, and climate impact analysis. In general, the studies include experimental designs (n=7), data analysis (n=4), systematic reviews (n=3), and simulations (n=1). The experimental and data analysis studies employed innovative techniques such as machine learning (n=4) and computer simulations (n=3). The investigations were carried out in different regions, highlighting Asia (India, China, Malaysia, Singapore; n=8), America (United States, Ecuador, Peru; n=5), Europe (Spain; n=1), and Africa (Nigeria; n=1).

4. Discussions

The results indicate that the use of advanced predictive models, such as the Adversarial Autoencoder (AAE) and the Maximum Entropy Model (MaxEnt), is particularly relevant for forecasting crop yields across diverse regions and climatic conditions, as demonstrated by studies conducted in Singapore and China. These approaches allow for the integration of climatic, ecological, and socioeconomic data, improving the accuracy of predictions by accounting for multiple and complex factors affecting agricultural productivity.

In particular, the study from Singapore [1] highlights the capability of machine vision-processed images of romaine lettuce to predict crop yields with high precision. Model validation using the MSSIM index (ranging from 0.5 to 0.6) demonstrates the effectiveness of smartphones and OpenCV as accessible and reliable tools for agricultural analysis. This advancement in the use of low-cost technological tools underscores the importance of democratizing access to technology for small- and medium-scale farmers, particularly in countries with limited technological infrastructure.

As for the study in China [2], the Maximum Entropy Model (MaxEnt) appears promising for optimizing large-scale planting structures, with applications in areas such as the Naoli River basin. Here, the use of climatic and ecological data effectively predicted crop suitability based on climate variations and soil conditions. The validity of the model, evaluated through the Area Under the Curve (AUC) with values exceeding 0.8, underscores MaxEnt’s reliability, making it a trustworthy tool for agricultural decision-making in changing environments.

Simulation models used in Australia [3], on the other hand, focus on irrigated agriculture. By integrating data on costs, crop requirements, and climatic variability, these models not only predict crop yields under different climatic scenarios but also optimize water resource use, which is essential in water-scarce regions. The validation of these models by comparison with real data ensures that the results are representative and useful for farmers’ decision-making processes.

Moreover, the integration of IoT and machine learning, as observed in the study conducted in Ecuador [11], offers an innovative perspective by enabling real-time monitoring of critical environmental factors such as soil moisture, temperature, and pests. This not only facilitates accurate crop yield prediction but also promotes precision agriculture by optimizing resource use and improving agricultural sustainability.

However, despite advances in implementing technologies such as machine learning and simulation models, certain challenges persist. The heterogeneity of data used in the studies and the lack of global standards for implementing these models in different agricultural contexts may limit the applicability of the results. Additionally, the quality of climatic and agricultural data can affect prediction reliability. In this regard, further studies should focus on standardizing data collection methods and improving model validation techniques to ensure accuracy under diverse geographical and climatic conditions.

The results presented in the bar chart of the distribution of studies on predictive models in sustainable agriculture highlight the predominance of predictive models in studies on sustainable agriculture. This underlines the growing importance of data analysis-based approaches to anticipate future scenarios and optimize decision-making. The fact that five studies focus specifically on predictive models reinforces the perception that these tools are fundamental to address contemporary agricultural challenges, such as managing limited resources and improving sustainable productivity. Furthermore, the combined approach of historical and prospective perspectives, represented in three studies, reflects a growing interest in integrating knowledge from the past to foresee and plan for the future. Meanwhile, the review studies allow consolidating theoretical and practical advances in the field, providing a solid basis for future research.

On the other hand, the bar chart showing the use of emerging technologies, optimization techniques and statistical models evidences the massive incorporation of emerging technologies and advanced techniques in agriculture, marking a trend towards digitalization of the sector. Tools such as IoT, AI, Big Data and Blockchain not only demonstrate their ability to improve agricultural management and efficiency, but also reflect the rapid technological evolution of the sector. The prominence of optimization techniques, used to maximize efficiency in diverse agricultural contexts, highlights the need for dynamic and adaptive strategies that respond to changing conditions in the agricultural environment. Finally, the significant use of statistical models highlights the relevance of data-driven approaches for agricultural resilience, especially in climate risk management and decision-making in the face of variability in production.

5. Conclusions

In conclusion, predictive models not only help to understand the impact of climate change, but also become important allies to forecast scenarios and improve decision-making in the industry. Similarly, technologies such as artificial intelligence, IoT sensors, and big data are taking agricultural resilience to a new level, providing powerful tools to adapt to a changing environment. These innovations not only improve efficiency, but also open the door to more sustainable and smart practices. But the challenge remains to make these technologies accessible to all, especially those facing greater vulnerability. The future of permaculture will depend on our ability to ensure that these solutions are inclusive, equitable, and truly transformative.