Analyzing the Performance of Machine Learning in Weather Forecasting

Abstract

Forecasting storms with significant rainfall is a difficult task for meteorologists. Several Machine Learning (ML) models, including the Lasso regression, ridge regression, elastic net regression, random forest, gradient boosting, and the decision tree regress or, are examined here. Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Square Error (MSE), and R2 score have all been used to assess the quality of the models. Using the rainfall dataset, this study compares and contrasts several machine learning regression methods. After carefully considering the outcomes of six alternative ML models, we settled on Lasso regression of the linear model as the most effective. The Lasso model obtained a 99.21% R2 score, 13.68 MAE, 6432.41 MSE, and 80.20 RMSE when used on 80% of the training data set and 20% of the test dataset, respectively.