Filtering Outlier Data Using Box Whisker Plot Method for Fuzzy Time Series Rainfall Forecasting

Rainfall forecasting provides benefits in several sector. The pattern of rainfall intensity in the same month every year has similarities, so that modeling of fuzzy time series can be used to model rainfall pattern in a region, but the amount of rainfall in every month has a varied value which where there is too high rainfall values and too low (outlier). The value of the outlier can damage the error distribution causing the forecasting value to be not good, so it needs an outlier search method to optimize the fuzzy time series method. In this research has prposed the model used box whisker plot method to find outlier data and then compare the result fuzzy time series method with outlier data and data with outlier that have been omitted. The accuracy value is better indicated by the decrease in MAD value where the initial MAD value forecasting with outlier data is 114.39 and the predicted MAD value forecasting without outlier data is 93.85.