Dhoriva Urwatul Wutsqa

The application of Elman recurrent neural network model for forecasting consumer price index of education, recreation and sports in Yogyakarta

Recurrent neural network is a network which provides feedback connections. This network is believed to have a more powerful approach than the typical neural network for learning given data. The current research was aimed to apply the simplest recurrent neural network model, namely the Elman recurrent neural network (ERNN) model, to the consumer price index (CPI) of education, recreation, and sports data in Yogyakarta. The pattern of CPI data can be categorized as a function of time period, which tends to move upwards when the time period is increased, and jump at some points of the time period. This pattern was identified as similar to the pattern resulted by the function of the truncated polynomial spline regression model (TPSR). Hence, this research considered ERNN model which the inputs such as in the TPSR model were established by taking into account the location of the knot or jump points. In addition, the ERNN model with a single input, a time period was also generated. The results demonstrated that the two models have high accuracy both in training and testing data. More importantly, it was found that the first model is more appropriate than the second one in testing data.

Neuro fuzzy model with singular value decomposition for forecasting the number of train passengers in Yogyakarta

The neuro fuzzy model is a model that combines fuzzy and neural network, which has been applied to time series forecasting. A singular value decomposition method can be utilized for optimization of the neuro-fuzzy model based on the singular values of the matrix. This research aims to forecast the number of train passengers of PT Kereta Api Indonesia (Persero) Operating Region VI Yogyakarta by applying the neuro-fuzzy model with singular value decomposition. The forecasting accuracy of the proposed model is compared with those of the one order Takagi Sugeno Kang fuzzy model and the neuro-fuzzy whose optimization is done by the least square method. The results demonstrate that neuro-fuzzy models with singular value decomposition are more accurate than the other two models on testing data but not better on training data.

OPTIMIZATION OF NEURO FUZZY MODEL FOR DATA TIME SERIES WITH SINGULAR VALUE DECOMPOSITION METHOD

This study aims to develop new procedures in optimal neuro fuzzy modeling for time series data. Specifically in this research, the development of new procedure in modeling fuzzy Takagi-Sugeno-Kang order one for time series data which parameter-parametemic determination is done by singular value and neural network decomposition method, in order to obtain method of forming neuro fuzzy model for time series data optimal. In this research, we have developed a procedure to get the optimal Takagi-Sugeno-Kang fuzzy model for time series data by optimizing the parameter value search in consequence of fuzzy rule using singular value decomposition method. A new model of neuro fuzzy modeling is optimized, the fuzzy model whose parametem optimization is based on the neural network by the singular value decomposition method. Parameters in consequent part of the rule of fuzzy are optimized by the singular value decomposition method and the parameters in the antecedent part of the fuzzy rule are optimized based on neural network backpropagation with gradient descent method.