Implementation of local regression smoothing and fuzzy-grammatical evolution on rainfall forecasting for rice planting calendar

Preprocessing method is the important step to get the better result in learning process, especially ini time series forecasting. The appropriate preprocessing method will affect the results of the learning process. Local Regression Smoothing (LRS) is the one of preprocessing method used for extreme data or outliers in time series data. The result of LRS used for input data in learning stage used Fuzzy-Grammatical Evolution (GE) algorithm, which optimize GE algorithm in crossover and mutation parameter to forecast the rainfall data in Bandung regency from year 2006 to 2015. The experiment result obtain the accuracy of rainfall forecasting is 91.10% and then developed the rice planting calendar based on the rainfall forecasting have the accuracy 84.62%.