Missing Value Prediction for Qualitative Information Systems

Most information systems usually have some missing values due to unavailable data. Missing values have a negative impact on the quality of classification rules generated by data mining systems. They make it difficult to obtain useful information from the data set. Solving the missing data problem is a high priority in the fields of knowledge discovery and data mining. The main goal of this paper is to suggest a method for converting a qualitative information system into a binary system, by using a distance function between condition attributes, we can detect the missing values for decision attribute according to the smallest distance. Most common values can be used to solve the problem of repeated small distance for some cases. This method will be discussed in detail through a case study.