Research And Implementation Of Stock Forecasting Based On Improved K-means Algorithm

With the development of economy people pay more attention to the stock market and a lot of people are engaged in the stock market.The stock market forecast is difficulty due to the high complexity of the stock market and the more factors of affecting the stock market.Stock market experts have been committed to the study of the stock market and put forward a higher forecasting model to reveal the operation mode of the stock market,to provide investors with investment decisions.How to effectively use the historical investment data for data mining to provide investment recommendations for individual users is a problem that is worthy of study.With the constantly development of data mining technology,more and more financial time series data need to be processed for the financial industry.In the financial asset investment,individual users are always in high risk compared with business users.At present,most of the financial data mining algorithms that the study of data mining in individual user and recommendation of investment is less face to enterprise.Financial time series data of individual users have different characteristic compared with enterprise,and it can't utilize the traditional data mining algorithms.This paper uses the improved k-means algorithm to recommend the interested stocks with higher probability of rising for the users aiming at the investment behavior for individual users.Improving the traditional linear regression model and its parameter calculation,the paper proposed the PRA dimension reduction algorithm and reduced the original data characteristic to the parameter dimension of the linear regression model,which reduced the time complexity.It searched the similarity of data characteristic of the PRA reduced dimension and clustered the higher similarity data through the similarity search algorithm.The cluster in higher similarity was regarded as leaf node to construct the core tree.It iterated the core data to improve the branches of the core tree.Processing subsequently the leaf nodes by the k-means algorithm enhanced the robustness of the core tree.Based on the improved algorithm this paper implemented a financial data mining of individual users recommend system based on clustering algorithm for individual users.The experiment and test results showed that the system with improved algorithm could recommend the financial investment in real-time and give therecommended results and the score ranking of each result through financial time series data of the corresponding period the customer input effectively,the higher the rank,the higher the investment income of the stock or financial asset,which greatly reduced the investment risk of the individual customer and increased its investment income.Through the design and implementation of this system,we hope that lay the foundation for future research of the individual user investment recommendation system.