Rainfall Prediction using Regression Model

Weather forecasts are good benefits for society and sustainable development. Farmers need help to maximize growth efficiency, save resources, and optimize production. To achieve these things, the farmers need to have an accurate weather prediction system that will improve their decision-making on typical farming activities such as planting and irrigating. Prediction of rainfall is a hard work in climate forecasting manner. Machine learning techniques can predict the fall of rain by bringing out hidden forms from recorded weather data. Rainfall prediction refers to the knowledge of the weather-related parameters, which are temperature, pressure, humidity, wind speed, etc. This paper proposes the rainfall prediction system using ‘SoftMax’ or multinomial logistic regression. The system performs the prediction process upon the historical weather information and produces the one of the five categories of rainfall, together with the wind speed and temperature. In this rainfall prediction system, the weather data for eleven cities of Myanmar is currently acquired from ‘wunderground’ and applied in order to test the prediction process. The experiment shows that the system can predict 83% accuracy in ‘SoftMax’ logistic regression based rainfall prediction.