Comparison of LR, SVR, and DNN for the Rainfall Forecast of Guwahati, Assam

Abstract

Forecasting of rainfall is an important research area from a farmer’s perspective. This paper presents the comparative assessment of the linear regression (LR), support vector regression (SVR), and deep neural network (DNN) for the forecasting of rainfall for Guwahati. For the research, the daily rainfall of the Guwahati region is collected from Regional Meteorological Center, Guwahati, Assam, India. The daily rainfall data is predicted based on the value of time, minimum temperature, maximum temperature, humidity, STP, mean sea level, STN, and wind speed. The algorithms are compiled with the different ratios of the training and testing set of rainfall data. The mean absolute error (MAE), mean squared error (MSE), and root mean squared error (RMSE) are reported to compare the LR, SVR, and DNN. The DNN model performs well as compared to the LR and SVR based on MSE and RMSE, but the SVR model outperforms based on the value of MAE. Both models can be considered as the best model irrespective of the ratio of the train and test set of rainfall data.