Stock Market Prediction Based On Neural Network Data Mining Model

To the question of predication in the stock market,I study some informations about data minging predication model of stock price based on neural network.I put forward to do attribute relativity analysis for inputing index variables and outputing object variables with the theory of attribute relativity analysis in the course of designing a neural networ predication model.And I do a demonstration analysis based on history exchange data of stock market.Thereout,I hope to achieve the object of optimizing the inputing variables's combined mode and improving the predication capability of a network model.Roughly work flow as:the first step,calculate the index date of technique analysis with the technique of data-base query.the second step,generalize the index data and calculate the relativity between generalized inputing index data and outputing object with the measurement units of information entropy,select four kinds of index attributes whose relativity is bigger than others as inputing variables of network model.The last step,select two mutual-substantive data sets as training set and testing set,then establish sequential and discrete RBF neural network model,after some kind of training and testing,accomplish the object of predication of stock price.The technique of data generalization and attribute relativity analysis are also adaptive for other data minging models,such as decision-tree algorithm.