#multivariate linear regression

The main objective of this work is to analyze different atmospheric factors and their correlation to daily rainfall, and then further compare them using various machine learning models such as Multivariate Linear Regression (MLR), Decision Tree Regressor, Random Forest Regressor, XGBoost Regressor and our proposed Stacked LSTM RNN model for forecasting rainfall. Daily rainfall data for almost 40 years are taken as input for this study and comparison of the above mentioned models have been carried out with respect to the Mean Absolute Error (MAE), Coefficient of Regression (R2) and the Mean Squared Error (MSE). The LSTM model emerged as the top performer, displaying the lowest MAE score. Hence, it was chosen to forecast daily rainfall for 7 days, weekly rainfall for 4 weeks and monthly rainfall for 12 consecutive months. Notably, the model’s accuracy improves as the duration of recorded observations are increased. The key novelty presented in this work is the proposed stacked LSTM Model and the comparative study of the model to predict rainfall in daily, weekly and monthly format. Our study prominently underscores the effectiveness of the proposed stacked LSTM Model in accurately forecasting rainfall.

The main objective of this work is to analyze different atmospheric factors and their correlation to daily rainfall, and then further compare them using various machine learning models such as Multivariate Linear Regression (MLR), Decision Tree Regressor, Random Forest Regressor, XGBoost Regressor and our proposed Stacked LSTM RNN model for forecasting rainfall. Daily rainfall data for almost 40 years are taken as input for this study and comparison of the above mentioned models have been carried out with respect to the Mean Absolute Error (MAE), Coefficient of Regression (R2) and the Mean Squared Error (MSE). The LSTM model emerged as the top performer, displaying the lowest MAE score. Hence, it was chosen to forecast daily rainfall for 7 days, weekly rainfall for 4 weeks and monthly rainfall for 12 consecutive months. Notably, the model’s accuracy improves as the duration of recorded observations are increased. The key novelty presented in this work is the proposed stacked LSTM Model and the comparative study of the model to predict rainfall in daily, weekly and monthly format. Our study prominently underscores the effectiveness of the proposed stacked LSTM Model in accurately forecasting rainfall.