Predicting Stock Trends Using Tsk-fuzzy Rule Based System

Abstract

Stock investment is often regarded by investors as a green opportunity that is characterized by high rewards and its attendant risk. Consequently, investors are preoccupied with analysis and prediction of the future performance of stocks, and the direction and magnitude of future changes in the stock value. In this study, the Takagi-Sugueno-Kang - Fuzzy Rule Based System (TSK-FRBS) was used to analyzed the trend of stocks using historical data gathered over a period of five months between January and May 2015. The TSK-FRBS was implemented in R. The input data is split into training and testing data for experimentation, testing and further analysis. Predictive accuracy was evaluated using the root mean squared error (RMSE) and symmetric mean absolute percentage error (SMAPE). Final results showed consistency obtained from feeding the model with the data and hence proved that patterns that allow for prediction can be deduced from the chaotic nature of the stock exchange market.