A Case Based Clustering-Based TSK Fuzzy Rule Systems for Stock Price Forecasting

Stock price predictions suffer from two well known difficulties, i.e., the high dimensionality and non-stationary variations within the huge historic price data. This paper presents a new financial time series-forecasting model by a case based clustering TSK fuzzy rule system for stock price predictions in Taiwan Stock Exchange Corporation (TSEC) . This forecasting model integrates a case based reasoning technique, a TSK Fuzzy Rule based system (TSK), and Simulated Annealing (SA) to construct a perdition-system based on historical data and technical indexes. The model is major based on the idea that the historic price data base can be transformed into a smaller case-base together with a group of TSK fuzzy decision rules. As a result, the model can be more accurately react to the current price of the stock from these smaller case based TSK fuzzy rule system .MAPE is applied as a performance measure and the effectiveness of our proposed CBTSK model is demonstrated by experimentally compared with other approaches on various stocks in TSEC. The average MAPE of CBTSK model is 85% the highest among others.