Optimization of Dynamic Naive Bayesian Network Classifier with Continuous Attributes

摘要

Classification accuracy is the most important performance indicators of the classifier. Feature subset selection is an effective method to improve the classification accuracy of the classifier. Existing methods of feature subset selection focus on the static classifiers, lacking research on feature subset selection of dynamic classifier. This paper first proposes dynamic naive Bayesian network classifier with continuous attributes and the evaluation criteria of the dynamic classification accuracy. Basing on this, we establish feature subset selection optimization method of dynamic naive Bayesian network classifier; experiments and analysis are done using real macroeconomic time series data.