Trend Feature Mining Algorithm Based on Financial Time Series

In this paper, we transform shares time series into price rate of change (ROC) time series based on the characteristics of financial time series, improving the trend feature algorithm and clustering algorithm, and providing a new corresponding feature similarity measure, therefore we can forecast time series into the detection of frequent and effective sets, which can help us to make data mining forecasts. Experimental results indicate that the proposed method is effective in event forecasting.