The Analysis of Incomplete Business Data

Abstract

Missing values can dramatically reduce the accuracy and availability of missing data, especially when analyzing business data. A common method to deal with the missing data is simply deleting the samples containing missing attributes. However, this will lead to bias and invalid conclusions since some data are too important to be omitted easily. Therefore, we should use certain methods to complete the data set instead of deleting data with missing values. In this paper, we compared several data imputation methods by adopting them to deal with six benchmark business data sets. The result provides us with guidance when dealing with incomplete business data.