Research On Information Uncertainty Classification Decision Algorithm

With the development of information technology,a large amount of data information is stored in all walks of life,and the total amount of global data information is expanding rapidly.Due to the accuracy of machine equipment and the omission of personnel processing data,data information is mixed and error,missing,redundant or inconsistent.Data missing filling and uncertain information classification decision play an important role in data processing.For data sets containing a large number of samples,this paper studies the classification decision algorithm for information uncertainty data.The main research contents are as follows:In view of the problem of missing data,this paper first analyzes the situation and principle of missing data from the aspects of the causes and mechanisms of missing data,and introduces the existing methods to deal with missing data from different angles.The overall perspective is divided into three methods : deletion,filling and non-processing.The filling method based on the similarity between attributes is divided into linear regression filling and nonlinear regression filling.Based on the similarity between samples filling method is divided into mean filling and nearest neighbor filling method;and parameter estimation expectation maximization algorithm.In view of the above algorithm,an improved density clustering algorithm is proposed.The missing values are filled by selecting the conditional attributes of strong correlation coefficient,and the linear difference between strong correlation attributes and weak correlation attributes is analyzed.The filling accuracy of the algorithm and other algorithms and the classification decision accuracy after filling are evaluated by using wine category data sets in different missing rates.For the problem of data classification decision algorithm,the uncertainty data processing framework is analyzed,and the classification algorithm of approximate rough set and decision resolution is proposed.The algorithm principle is based on the decision tree modeling of approximate rough set and decision resolution.The approximation degree of rough set is used to judge the ability of attribute to divide sample data,and it is substituted into the decision resolution algorithm.The attribute with the largest decision resolution is used as the splitting feature to establish the classification decision tree.The sample attribute value is brought into the data classification decision tree,and the corresponding prediction results are obtained by traversing each layer of the tree.So as to improve the classification prediction accuracy and computational efficiency,reduce the tree complexity.In this paper,an improved density clustering algorithm is proposed to fill the sample set containing missing values,and a classification algorithm based on approximate rough set and decision resolution is proposed,which provides theoretical support and basis for sample classification prediction.