Outlier Detection Based on Surfeit Entropy for Large Scale Categorical Data Set

Number of methods based on classification, clustering, frequent patterns and statistics has been proposed to collect meaningful information by removing unwanted data. Information theory uses statistical approach to achieve its goal. The outlier detection from unsupervised data sets is more difficult task since there is no inherent measurement of distance between objects. Here in this work, we proposed a novel framework based on information theoretic measures for outlier detection in unsupervised data with the help of Max/Surfeit Entropy. In which we are using different information theoretic measures such as entropy and dual correlation. Using this model we proposed SEB-SP outlier detection algorithm which do not require any user defined parameter except input data. We have also used the concept of weighted entropy. Our method detects outliers better than existing approach.