Prediction of Southwest Monsoon Rainfall for Kerala, India using ε-SVR Model

The current study has presented an s-Support Vector Regression (E-SVR) based prediction model for the prediction of southwest monsoon rainfall for the Kerala state of India. There is a need for accurate prediction of the rainfall over the state as the state has suffered from floods and droughts in the recent past. The proposed E-SVR prediction model with combined optimal outcomes i.e. Support Vector Regression Optimal (SVRO) provides the goodness of fit (R2 = 0.96) value, whereas Ensemble of different SVR models (ESM2) by changing the meta parameter, epsilon provides the goodness of fit (R2 = 0.79) value. Thus, the proposed ε-SVR model is providing accurate predictions after setting the optimal meta parameter i.e ε-insensitive zone and can be used as a prediction tool for analyzing complex weather events.