An uncertain data model construction method based on nonparametric estimation

Abstract

Uncertain data may exist in many application fields, due to the inaccurate raw data, the use of coarse-grained data set, for the purposes of privacy protection, and the data integration etc. The original features of the data may be changed or ignored if the uncertainties were mishandled. Therefore the effective management and analysis of uncertain objects should rely on an appropriate uncertain data model depicting the characteristic of uncertainties. For the uncertainties in the values of data attributes, this paper proposed an uncertain data model construction method based on nonparametric estimation, which can represent the uncertainty distribution characteristic efficiently without assuming the data distribution. And the effectiveness of the proposed algorithm was verified through the experiments on UCI and real data sets.