| The linkage of stock prices,that is,the phenomenon of “same rise and fall” that often occurs in the stock market,means that in a certain period of time,a certain type of stock that belongs to a certain fundamental category or has a certain conceptual factor is presented.The simultaneous rise or fall of the stock price has caused some stocks in the stock market to rise and fall.With the increase in the number of listed companies in China and the continuous expansion of business operations,there are more and more standards for the division of industries,and the links between stocks represented by various listed companies are becoming more and more complicated.Such correlation research helps to rationally price stocks in different markets,and the linkage rules between stocks also have a strong reference for futures trading and hedging strategies of wealth managers.Therefore,the judgment of the correlation between each stock and its market trend has extremely important application value and theoretical significance.This paper mainly uses the association rule algorithm to analyze the stock market data.As the most classic Apriori algorithm in the association rule algorithm,it generally deals with discrete data.It can’t directly use the Apriori algorithm to directly analyze the continuously changing data of stock price.The traditional stock data-based mining is to roughly divide the stock price into two categories: “rise” and “fall”.This correlation does not describe the true linkage relationship of stock prices very well,especially when the support and confidence thresholds are too low.At the same time,a large number of redundant and similar association rules will be generated,which makes the understanding and practical application of the association rules more difficult.Therefore,based on the Apriori algorithm,two optimization algorithms for setting the threshold are proposed,which are the association rule algorithm based on the rise and fall,the number of days threshold and the association algorithm based on the interval threshold.Applying to the correlation analysis mining of Shanghai and Shenzhen A-share stock market data,we can find that the two optimization algorithms have a very good processing effect on the continuous data of stock price changes.Overall,the main work of this paper includes:(1)Using reptile technology to collect stock trading data,using Python software,through the TuShare financial data interface package,obtained a total of 100 days of Shanghai and Shenzhen A during the market opening period from September 29,2017 to March 02,2018.The stock market data is stored in the MySQL database to meet the subsequent stock analysis research.(2)According to the characteristics of stock data,using some data preprocessing methods,the continuous numerical data is converted into a Boolean transactional data set,which is transformed into a “shopping basket” model,so that the classic data mining Apriori algorithm can be used to mine between stocks.Association rules.At the same time,for the continuously changing data,the Apriori algorithm based on the rise and fall and the threshold of days is proposed.The setting of the rise and fall threshold and the threshold of the day is added,and the Apriori algorithm is optimized before the frequent itemsets are generated.Using the Shanghai and Shenzhen A-share stock market data stored in the Mysql database,the classical Apriori algorithm and the Apriori algorithm after adding the threshold are experimentally verified.It can be seen that the association rule generated by the Apriori algorithm after adding the threshold reduces the frequent items.Based on the set,saving on the running time,the confidence and support are improved.The validity and feasibility of the algorithm are proved.(3)Aiming at the limitations of the degree of promotion and confidence setting in association rules,an algorithm based on interval threshold setting is proposed.N intervals are set between ±10% of daily stock price fluctuations,and daily stock price increase and decrease data is used.The stocks are allocated to the corresponding intervals,and the higher-priced stock codes that are simultaneously placed in the same interval are extracted,and the algorithm is applied to the Shanghai and Shenzhen A-share stock market data,and the financial insurance industry,the real estate industry,and the cultural media are mined.Stock association rules in the industry,food and beverage,etc.(4)Through the data visualization technology,the chord diagram and the association rule causal map are drawn to visually show the relationship between stocks,which further validates the validity and feasibility of the algorithm based on the rise and fall,the number of days threshold algorithm and the interval threshold setting algorithm. |
PDF Full Text Request