Forecasting Weekly Rainfall Using Data Mining Technologies

Rainfall forecasting is a technologically and scientifically a challenging task around the world. Rainfall is one of the most important weather conditions in a given area. Forecasting possible rainfall can help to solve several problems related to the tourism industry, natural disaster management, agricultural industry etc. As the Sri Lankan rural economy is mostly based on agriculture, it is important to forecast rainfall as well as other weather conditions accurately. The weather patterns are localized and hence, generalization of weather prediction models is very difficult. Therefore, this project proposes three data mining models to forecast rainfall, and compares the prediction performances of those models. To that end the data mining models linear regression, SMO regression, and M5P model were trained from rainfall data collected from the Badulla district, Sri Lanka, during the period 2002 to 2017, to forecast weekly rainfall for the following five months lead-time. Each model was evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Root Relative Squared Error (RRSE), Root Absolute Error (RAE), Direction Accuracy (DA) and residual analysis. According to the findings, the M5P model tree provided the lowest error value, highest direction accuracy, highest correlation between actual and predicted rainfall values, and better randomness of the error values compared to the linear and SMO regression models