Distribution-based selective classifiers for incomplete data

Selective classifiers are a kind of algorithms that can effectively improve the accuracy and efficiency of classification by deleting irrelevant or redundant attributes of a data set. Due to the complexity of processing incomplete data, however, most of them deal with complete data. Yet actual data are often incomplete and have many redundant or irrelevant attributes, a selective classifier for incomplete data (SDBNB), which is based on a newly constructed Bayes classifier (DBNB), is presented. Experiments results from twelve benchmark incomplete data sets show that the average accuracy of SDBNB is 0.69 percent and 0.58 percent higher than that of the effective selective classifiers: SNB and SRBC. Furthermore, its standard deviation is 0.11 and 0.05 lower than that of SNB and SRBC. This shows that not only SDBNB has higher accuracy, but also performs more stably as well.