G probability-based method and its application

In this paper, we introduce a novel G probability-based classification algorithm (GPC) and apply it on real stock market data to predict the ´upward´ or ´downward´ trend of stock returns one day in the future. To improve the prediction accuracy we use the discrete Fourier transform and its inverse transform to filter noise and conduct data reduction whilst preserving its overall movement in the time domain. The experimental result shows that the GPC algorithm obtains higher average classification accuracy on 15 chosen stocks time series than kNN and naive predictor which have been implemented in our stock analysis and prediction prototype (SA&P) for comparison purpose.