Forecasting shanghai composite index based on fuzzy time series and improved C-fuzzy decision trees

Abstract

Followed with Song and Chissom’s fuzzy time series model, many fuzzy time series models have been proposed for forecasting combined with some technologies or theories. This study presents a new forecast model on basis of fuzzy time series and improved C-fuzzy decision trees for forecasting stock index which is one of the most interesting issues for researchers. There are two main improvements for C-fuzzy decision trees in this paper. The first one is that a new stop condition is introduced to reduce the computational cost. The other one is fuzzy clustering with weight distance computed with information gain. And then weighted C-fuzzy decision tree (WCDT), a novel forecast model armed with k nearest neighbors, has been proposed and experimented on Shanghai Composite Index over a ten-year period, from 1997 to 2006. The empirical analysis not only demonstrates the forecasting procedure and the way to obtain the suitable parameters, but also shows that the proposed model significantly outperforms the conventional counterparts.