Determining the time period and amount of training data for Stock Exchange of Thailand index prediction

This paper analyzed the time period and amount of training data that were used in applying the prediction function to forecast the Stock Exchange of Thailand index (SET Index). The training data were grouped into one year and two years over the period of 2003-2008, and the testing data were collected from January 2005 to March 2009. This paper compared the results of using different time periods training data in order to And the best training data set that will most accurately predict the SET index. The results show that there is no significant difference between using one year and two years training data with MAPE lower than 1%.