Processing the Missing Value Based on the Linear Regression Approach

The problem of missing data is quite common, so solving the problem of missing values is necessary to greatly improve our data mining and analysis problems correctly. Currently, there are many different approaches to dealing with missing values or often called missing data, each of which has different advantages and disadvantages, the main drawback is that it doesn't care about the effect of the data input for the attribute. In this paper, we use the linear regression method in statistical theory for processing. The advantage of the method that does not alter the original data is an important issue in data preprocessing. The implementation method of this approach is based on the affinity matrix between attributes, find the attributes with the largest affinity values, then build a linear regression equation, from which the values will be filled in. missing value for attribute.