Population growth and increased migration from rural to urban areas have led to widespread climate change that has significantly impacted land use. Urban sprawl is a phenomenon that happened these years, especially in developing countries. Therefore, planners have always been looking for methods and models that simulate the expansion of urban and climatic land use well to prevent the unbalanced growth of cities, climates, and undesirable development problems. These models guide them to manage the plans in the desired direction. Advances in artificial intelligence in recent years, along with widespread access to online data, the emergence of new methods of big data analysis, and the development of advanced technologies, have led to the emergence of new technologies and methods such as machine learning techniques and agent-based modeling. Investigating Iran's policies on land use issues and developing new solutions with considering a comprehensive review of data-driven methods are needed to analyze the problems and solve the problems resulting from these changes. In addition to data-driven approaches, the specific benefits of factor-based models include their ability to model individual decision-making institutions and their interactions, the combination of social processes and non-monetary influences on decision-making, and the dynamic linkage of social and environmental processes. Therefore, classification, forecasting, modeling, and simulation to estimate the future situation with the help of data from these changes in different periods can be the basis for making the right decisions in the current situation. In this regard, experts in this field have always considered the use of new strategies for land modeling and land use planning. Although extensive studies have been conducted in the field of machine learning (ML) methods as a new approach to classification, prediction, simulation, and modeling in various fields of science; However, these studies have less reviewed the proposed and applied methods of the agent-base modeling and machine learning in the analysis and modeling of land-use change studies. To this end, this article provides the opportunity for a systematic review of the application of machine learning algorithms and agent-based modeling, which has been recorded in the most critical research and experimental evidence of the United States, Europe, and various parts of Asia, especially East Asia and also Iran. Therefore, the different algorithms and methods implemented in each study are reviewed, and the results of data analysis are presented accordingly, which can be the basis for further research to use widely used, accurate and dynamic models. This study shows that different land use issues such as classification, forecasting, and simulation require algorithms with appropriate structure. Results show that no method and algorithm can be considered absolutely superior compared to other methods and algorithms. Thus, the most widely used methods for classifying, predicting, and simulating land-use change are categorized in this paper. In general, it was also found that support vector machine (SVM) and Convolution Neural Networks (CNN) as widely used methods, with the best results, provide valuable solutions for land use classification, forecasting, and simulation.