Implementation of Hybrid RNN-ANFIS on Forecasting Jakarta Islamic Index

Abstract

RNN is a type of artificial neural network used to handle problems that require sequential data processing. ANFIS is a method that combines the advantages of fuzzy logic and artificial neural networks to create a system, so can adapt the parameters it uses according to the obtained data so that it can build an automated inference system. In this research, we make combination of RNN in ANFIS, which makes ANFIS able to accept input in the form of time series data so that ANFIS can recognize patterns contained in the time series data and its suitable for forecasting cases in the Jakarta Islamic Index. The membership functions used are three Gaussian functions. The results of the RNN-ANFIS Hybrid model training provide an interpretation that the first membership function represents the trend change indicator value, the second membership function represents the closing price change value in the last eight days, and the third membership function represents the pattern change value in the trend. The model for the Jakarta Islamic Index provides very good predictions with an MSE value of 0.001 and an MAE of 0.246.

Keywords

• rnn-anfis
• neural network
• fuzzy logic
• jakarta islamic index