| The stock market of our country is developing rapidly and to be mature. With thedevelopment of the stock market, the research of trend of stock curves trend still hasvery strong application value. Technical analysis, as an important part of securitiesanalysis, the research at home and abroad has reached a higher level. With thedevelopment of information technology, new theories and technologies of analysismethod has been applied to the analysis of technology constantly. At present, thereare many types of stock analysis methods in the market, but all have their ownadvantages and disadvantages, or have a variety of problems. Through the analysis wefound that for many stock prediction analysis, there are problems such as complexoperation process, not accurate or not easy to quantitative analysis.This paper mainly studies that data mining in the trend of the stock price curvespredict. The paper starts from the listed company’s historical stock data, applies somecommon concepts and methods in data mining to analyze its future stock price inorder to predict the future of the company’s share price trend, and compares with itsactual trends, to analyze its applicability. The main contents include the followingaspects:The paper first describes basic techniques in data mining and securities analysisand demonstrates the applicability of data mining in the prediction of stock markettrends. Based on this, it researches some existing methods, including time-seriesmethods and Markov methods, and experiments. For time series prediction, it usessingle moving average method, single exponential smoothing method and secondexponential smoothing method. According to the results, we can find that singleexponential smoothing method avoids the problem that sample data cannot be coveredand cannot be judged by data’s influence in single moving average method andsecond exponential smoothing method avoids the problem that single exponential smoothing method is only suitable for problems with horizontal trends in time-seriesanalysis. According to the improvement strategies to avoid defects in the abovemethod, it analyzes the experimental results Markov forecast method, found itsdisadvantage and Markov forecasting method considering the effect of volume ispresented, and then experiment to verify it. Finally it sums up all the methods andproposes how to prevent defects in new method for the state may not exist, to improvethe efficiency.We can see from the experiment: time series prediction is easy to calculate andhas simple process, but has many errors; Markov forecast method has higher accuracy,but it only consider the impact of a single factor, so there is still improvement room;Markov forecasting method considering the effect of volume take into account theimpact of different factors, so it’s most accurate, suitable for prediction of single stockshort term price trends.The methods used in this paper are relatively easy to quantify the realization, andthe results are clear, so it has a certain reference value for actual trend prediction ofthe stock. |
