Optimization weather parameters influencing rainfall prediction using Adaptive Network-Based Fuzzy Inference Systems (ANFIS) and linier regression

This paper conducted a study to investigate the ability of Adaptive Network-Based Fuzzy Inference System (ANFIS) in doing modeling to determine the weather parameters that influence the output parameters of rainfall (RF) and have good predictive ability. Plotting the data of the prediction is also made to the Linear Regression (LR). The data is tested daily at the weather station in Badau area, Belitung province, Indonesia. A total consisting of 433 pairs of data for 1 year containing seven weather parameters as input and one parameter as output. As for the performance evaluation criteria used indicator of the ability of ANFIS statistic model: Pearson correlation coefficient (r), coefficient of determination (R2) and root mean squared error (RMSE), from several input parameters in the analysis, 1-input RHmax most optimal influencing rainfall (RF) output, (RMSE = 1.8896 mm / day at the training phase and RMSE = 3.2370 mm / day at the checking phase). Plot the data ANFIS against Linear Regression, 1-input parameter RHmax has optimal value of the influence of rainfall (RF) output with optimal statistical indicator (R² = 0.7065, r = 0.8405, RMSE = 0.8732 mm / day).