A Data Cleaning Method Based on Association Rules

Abstract

The quality of the data affects the usability of the data mining’s results. Making a data preparation before the mining can improve the quality. If the data are collected from the multi-data source, data preparation becomes very difficult. In this paper, a data-cleaning method based on the association rules is proposed. The new method adjusts the basic business rules provided by the experts with association rules mined from multi-data sources, and generates the advanced business rules for every data source. Using this method, time is saved and the accuracy of the data cleaning is improved.