Forecasting the stock exchange of Thailand using data mining techniques

The stock price index movement is regarded as a challenging task of financial time-series prediction. An accurate forecasting of stock price movement may yield profits for investors. Due to the complexity of stock market data, predicting it is very difficult. This study attempted to develop three efficient predictive models and compared their performances in the daily stock exchange market of Thailand (SET). These models are based on three classification techniques: the uses of linear regression, decision trees, and artificial neural networks (ANN). Thirteen technical indicators were selected as inputs for the proposed models. Three comprehensive parameter settings in the experiments were performed. Experimental results showed that average performance of the ANN model (89.79%) was found to be significantly better than that of the linear regression (89.74%) and decision tree models (88.07%). Consequentially, this research demonstrates rule extraction as a post-processing technique for improving prediction accuracy and for explaining the logic to financial decision makers.