Authors : Agus Maman Abadi, Nurhayadi and Musthofa
Abstract: Jakarta Composite Index (JCI) is an indicator for monitoring the movement of the prices of all shares listed on the Jakarta stock exchange. Most studies of JCI prediction are conducted using conventional statistical methods. In this study, we come up with a new procedure to construct wavelet weighted fuzzy model and apply it to predict JCI. Wavelet weighted fuzzy modeling procedure is begun with wavelet transformation using a Discrete Wavelet Transform (DWT) mother Haar for time series data. The DWT results are used as an input of Mamdani Fuzzy Model. Furthermore, the weight of fuzzy rules is determined based on the training data. Finally, defuzzification process is performed to obtain the output of Wavelet Weighted Fuzzy Model. This procedure was applied to predict the value of JCI. The results indicate that the Wavelet Weighted Fuzzy Model has high accuracy for training and testing data. In addition, the prediction of JCI value is also performed with other models such as Weighted Fuzzy Model and Wavelet Fuzzy Model. Compared to the other models, Wavelet Weighted Fuzzy Model gives better results than that of the other models.
Agus Maman Abadi, Nurhayadi and Musthofa , 2017. Optimization of Wavelet Weighted Fuzzy Model for Time Series Data and its Application to Forecast Jakarta Composite Index. Journal of Engineering and Applied Sciences, 12: 5672-5678.