TAIEX forecasting based on fuzzy time series, particle swarm optimization techniques and support vector machines

• A new method for forecasting the TAIEX is presented.
• It forecasts the TAIEX based on fuzzy time series, particle swarm optimization techniques and support vector machines.
• The particle swarm optimization techniques are used to get optimal intervals in the universe of discourse.
• The support vector machine is used to classify the training data set.
• The experimental results show that the proposed method outperforms the existing methods for forecasting the TAIEX.

In this paper, a new method for forecasting the TAIEX is presented based on fuzzy time series, particle swarm optimization techniques and support vector machines. The proposed method to forecast the TAIEX is based on the slope of one-day variation of the TAIEX and the slope of two-days average variation of the TAIEX. Because the slope of two-days average variation of the TAIEX is smoother than the slope of one-day variation of the TAIEX, it is chosen to define the universe of discourse. The particle swarm optimization techniques are used to get optimal intervals in the universe of discourse. The support vector machine is used to classify the training data set. The first feature and the second feature of the support vector machine are the slope of one-day variation and the slope of two-days average variation of the TAIEX, respectively. The experimental results show that the proposed method outperforms the existing methods for forecasting the TAIEX.