Stock Sequence Pattern Mining Method Based on SWI-GSP Algorithm
Huachun Liu, Hua Du · 2017
In the mining of stock data association, investors are more concerned with such association rules as X (t1)→Y (t2), That is, the X shares rise on t1, and the Y shares rise at a certain probability on day t2. As such association rules can't be directly used GSP algorithm, for this reason, the GSP algorithm is depth analyzed and improved, and the SWI-GSP (sliding window interval) algorithm is proposed. In the SWI-GSP algorithm, such association rule: X (t1)→Y (t2) (N= t2-t1) is implemented by adding the time interval parameter N (N= 0,1,2,3··· ···). The sliding time window W is designed, and the time interval N is counted in W, and the sliding window W moves along the time axis of the stock trading to realize the correlation counting in a transaction. Through the Chinese A-share data for the sample experiment, experiments show that, the SWI-GSP algorithm is better mining sequential pattern with time interval than GSP.