Application Of Covering Algorithm In Stock Prediction

The securities market is the place of issuance and trading of stocks, bonds, funds and other portfolio. The issuance and trading of stocks greatly promoted the development of market economy. Enterprise do their financing by issuing stocks to address the shortage of funds in their development. It plays an active role in the rapid and healthy development of enterprises. On the other hand, for capital appreciation, investors must find investment. In the stock market they gain a substantial income by buying stocks as investment. But high-yield in the stock market is associated with high risk. The stock price is high or low not only depends on operation status and profitability of the company, but also changes because of influence of domestic and international economic, political and market sentiment. The stock present trend is not regular, how to predict the trend of stock price movement for gaining higher revenue put forward a new research topic.In recent years, many domestic and abroad scholars has established some stock prediction model and has achieved good results. Because of the large amount of stock data, the current algorithm model which valid only for little data is helpless. In this paper, we analyze and study the stock features samples by using covering algorithm, the main results are as follows:Discussing the significance of the stock forecast and prediction, outlining the theoretical methods of current stock forecast and analysis, then we propose a specific research approach.According to the definition of time series, introducing the m days cross-time series, then processing sample pretreatment. The m days cross-time series stock samples are trained and predicted by using BP neural network. At last We analysed the experimental results. The principles of covering algorithm are introduced in detail, and the shortcomings of field covering algorithm are analysed. We used genetic algorithm in sample optimization selection, and we predict Shanghai A Stock time series by using original covering algorithm and covering algorithm which has been improved, then the results of the two algorithms were compared and analysed.Shanghai A stock Wanwei Gaoxin is predicted by improved and original covering algorithm, experimental results show that the former prediction accuracy is significantly higher than the latter. The sample data must be projected to unit ball surface, however stock price is dynamic, the minimum and maximum are difficult to determine, so it is difficult to normalize stock data, and bring some problems to the network training this problem must be further explored and improved.