Performance Analysis and Evaluation of Machine Learning Algorithms in Rainfall Prediction

Abstract

Massive rainfall forecast is a significant problem for the meteorological department. This paper investigates the performance of the various Machine Learning (ML) models, namely Lasso regression, ridge regression, elastic net regression, random forest, gradient boosting and decision tree regressor. Those models performances have been calculated through the evaluation metrics such as R^2 score, Mean Absolute Error (MAE), Mean Square Error (MSE), and Root Mean Square Error (RMSE).  The objective of this study is to compare different machine learning regression algorithms in rainfall dataset. In this analysis, we conclude that the Lasso regression of the linear model is the best model among six ML models. Lasso model given more R^2 score is 99.21%, MAE is 13.68, MSE is 6432.41 and RMSE is 80.20 at 80 % training data set and 20% at test dataset.

Keywords: Rainfall forecasting, machine learning algorithms, R^2 score, MAE, MSE, RMSE.