New multivariate kernel density estimator for uncertain data classification

Abstract

Uncertainty in data occurs in diverse applications due to measurement errors, data incompleteness, and multiple repeated measurements. Several classifers for uncertain data have been developed to tackle this uncertainty. However, the existing classifers do not consider the dependencies among uncertain features, even though this dependency has a critical efect on classifcation accuracy. Therefore, we propose a new Bayesian classifcation model that considers the correlation among uncertain features. To handle the uncertainty of data, new multivariate kernel density estimators are developed to estimate the class conditional probability density function of categorical, continuous, and mixed uncertain data. Experimental results with simulated data and real-life data sets show that the proposed approach is better than the existing approaches for classifcation of uncertain data in terms of classifcation accuracy.