The Application Of Data Mining Technique In The Field Of Stock Analysis

In the late 1980s, data mining, as a new research area ,appeared quietly. The research purpose of data mining is that in large scale data concentrated, find the certain regularity information that hidden, people are interested in. Along with the development of data mining, this technology has been applied in business management, government offices, scientific research, project development and many other areas. This paper concerns an important field of the data mining: the application of association rules in stock.At first, this paper has done a general introduction for data mining and stocks, including the concepts of data mining, functions, steps, stock background, basic knowledge, analytical methods of stock and related software, as well as the study status at home and abroad. Then, introduce the focus of the paper in detail: classical algorithm -Apriori algorithm of association rules.By analyzing the two defects of Apriori algorithm:repeatedly scan the database and generating a large number of candidate sets, introduce Apriori_TIDS algorithm to improve mining efficiency, for the particularity of stock information, describe the problem of Apriori _TIDS algorithm in the stock data mining:Aproiri_TIDS algorithm in data pre-processing and mining process will result in the loss of valid data and can not be effective analysis combined the importance of stocks.Then,the core of the paper, propose the Aproiri_TIDS optimization algorithm based on the weight parameters from new and old data, the importance of stock, the user's interest. After the stock code, transaction time, inflation fell and the data are pre-processed, on the optimization Aproiri_TIDS algorithm do experimental verification of the stocks, experimental results show that mining association rules of the improved algorithm is more effective.Although the algorithm will be lower than the original Apriori_TIDS algorithm after introducing weighting parameters in efficiency, combining the timeliness of the user is not in a very high demand, the improved algorithm can satisfy the user's requirements, the most important is that it can provide more comprehensive and valuable, user-interest rules to support customer decision-making.