COMPARISON OF FUZZY TIME SERIES MARKOV CHAIN AND AVERAGE BASED FUZZY TIME SERIES MARKOV CHAIN IN FORECASTING COMPOSITE STOCK PRICE INDEX

Abstract

JCI's movement is a reference for investors to make decisions on whether to sell, hold, or buy shares. In this study, the Fuzzy Time Series Markov Chain (FTS-MC) and Average Based Fuzzy Time Series Markov Chain (Average Based FTS-MC) methods will be compared in forecasting the closing price of the JCI. These two methods have differences in determining the length of the interval where the length of this interval is very influential in the formation of Fuzzy Logical Relationships (FLR). As a result, the length of the interval will affect the forecasting result of the closing price of the JCI. The data used in the study was the closing price of the JCI from January 2016 to April 2022. The accuracy level of these two methods is viewed based on the Mean Absolute Percentage Error (MAPE) value. The test results showed that the Average Based FTS-MC had a smaller MAPE value, which was 1.590415%. In other words, the accuracy rate of forecasting the closing price of the JCI with the Average Based FTS-MC Method has an excellent accuracy rate of 98.41%.

Keywords:  Average Based, Fuzzy Logical Relationship, Fuzzy Time Series Markov Chain, MAPE