Adaptive Neuro Fuzzy Inference System (ANFIS) approach for modeling paddy production data in Central Java

Abstract

The aim of this research is to develop the procedure of constructing an adaptive neuro-fuzzy inference system (ANFIS) model for time series data. The procedure of development applies statistical inference for optimizing ANFIS architecture. In this study, the procedure of Lagrange multiplier (LM) test is used for selecting input variables. Firstly, several lags which are indicated significantly different to zero are divided into 2 clusters, and these lags are selected as optimal inputs of ANFIS based on LM test. Secondly, the cluster numbers of inputs are also determined by using LM-test procedure. Based on this result, a number of rule-bases are generated. The developed model is applied for forecasting paddy production data in Central Java. This study concluded that lag-1, lag-2 and lag-5 with 2 clusters are selected as the optimal inputs. The 1-1 and 2-2 rules are selected as optimal rules. Finally, the model can work well, and generates a very satisfying result in forecasting paddy production data. Based on the root mean squares error (RMSE) and mean absolute percentage error (MAPE) values, the ANFIS performance is better than performance of Autoregressive Integrated Moving Average (ARIMA) for forecasting.