Forecast Of Stock Price Movement In Weeks Based On Data-ming

With the development of national economy and unceasing enhancement of people's investment consciousness, stock become an important investment tool.Therefore, how to predict the stock movements reasonably to avoid the stock market risk become a hot issue in recent years. Many investors and academic researchers from multiple aspects, and propose a variety of methods, to forecast and analysis the trend of stock price fluctuation.This paper first extracted daily closing price data from Shanghai Composite Index which has full data each week. Then change the first three day's daily closing price data to a percentage rise and fall. Using the same means to the last three day's daily closing price data. We do conceptual classification to the scatter diagram. Research the associations among various states of the first three days and the last three days, the associations among various states of the last three days and the first three days of next week, associations in weeks and between weeks. Found that most of the state transition is sideways to sideways for ups and downs. And the continuity of the stock transfer is very strong, jumping to convert between states occupy a relatively small proportion. What make the bull market and bear market of stock is that a minority status' s jumping transfer, continuous transfer to stronger state or continuous transfer to weaker state. The results have certain reference value for stock's short-term prediction.Last, using the theory of minimum spanning tree, we divide nodes of tree into three categories, including main center node, sub-center node and outlier node, apply this idea to the industry sector of stock market and make an explanation of the market characteristics of these three types of sectors. Through analysis of these three sectors of the stock market we can obtain the following conclusions. The main center sectors which have a relatively stable price fluctuation represent the trend of a majority of sectors in the market. While the outlier ones which have a relatively active price fluctuation reflect the deviation of abnormal sectors. We choose one sector from central and outlier sectorsrespectively, using association rule mining the associations among various states of the first three days and the last three days. We found that significant differences in the state's transfer in one week that influence the diffident fluctuation between central and outlier sectors.