A Machine Learning Approach for Predicting Temperature and Precipitation Patterns

Climate modeling is one of the landmark important topics. This research explores the integration of machine learning techniques into climate modeling, aiming to develop a simplified model for predicting temperature and precipitation based on location and time. It begins with an analysis of existing climate classification systems and the potential for machine learning to enhance predictive capabilities. In this paper, climate data were obtained and processed, and different features were used for experimentations to deploy artificial neural networks. After various settings with different features were experimented, the final model exhibited improved accuracy with mean absolute error (MAE) decline was given. According to results, one hidden layer with two neuron network yields 2.04 and 30.12 errors, and one hidden layer with three neuron network yields 2.23 and 30.51 errors in terms of MAE metric.