Research And Implementation Of Predicting Missing Value Algorithms In Data Mining

Data Mining is widely applied to all areas of society. The success and applicability of data mining depend largely on the quality of data, but the issue of missing value is still unavoidable during the process of data acquisition. Because missing value affects the quality of data, missing value prediction is an important task in data preprocessing and a key step to improve the quality of data.This paper analyzes and studies GM(1,1) predicting missing value algorithm and MVC predicting missing value algorithm deeply. When predicting missing value, GM(1,1) predicting algorithm is necessary to simultaneously build every grey model for all sequences of data. But it can't make use of the correlation of the sequences of data and solve complicated and nonlinear relationships, which makes the predicting precision less accurate. MVC predicting algorithm can predict missing value by association rules. Not getting enough available association rules to predict all missing values, this predicting algorithm can not deeply resolve the problem of the filling rate and correctness.GM(1,1) predicting missing value algorithm is made improvements in this paper. I combines GM(1,1) model with three-layer BP neural network, and then develops gray neural network combined predicting missing value algorithm. This combined predicting missing value algorithm not only considers the correlation of data of each sequence, but also considers the correlation of the sequences of data. At the same time, it reduces the complexity of operations and improves the predicting precision of missing value. So this algorithm is better than single predicting model.MVC predicting missing value algorithm is made improvements in this paper. By introducing three-layer BP neural network, I develops MVC neural network combined predicting missing value algorithm. The combined predicting missing value algorithm takes full advantages of association rules and BP neural network and considers the precision of completing missing values, which improves the filling rate and correctness of missing values.