Comparative analysis of different rainfall prediction models: A case study of Aligarh City, India

Abstract

This research paper delves into creating and comparing rainfall prediction models, employing diverse machine learning algorithms, including Logistic Regression, Decision Tree Classifier, Multi-Layer Perceptron classifier (neural network), and Random Forest. The study aims not only to predict rainfall patterns but also to evaluate the performance of each model through metrics such as Accuracy, Cohen's kappa coefficient, and Receiver Operating Characteristic (ROC) curve analysis. Additionally, the relevance of the predictors employed in each model is thoroughly assessed. The results of extensive experimentation and analysis reveal that the Logistic Regression (Accuracy = 82.80 %, ROC = 82.45 %, Cohen's Kappa = 65.05 %) and Neural Network model (Accuracy = 82.59 %, ROC = 81.94 %, Cohen's Kappa = 64.40 %) has emerged as the most promising approach, achieving the highest percentage of accuracy, ROC and Cohen's Kappa metrics; among the models considered. This outcome underscores the effectiveness of Logistic Regression and Neural Network architectures in capturing intricate patterns and relationships within rainfall data.

Keywords

• Rainfall prediction
• Deep learning
• Machine learning
• Water resources