A data completation algorithm for incomplete continuous data set

Data completation is necessary in many practical applications. For the incomplete data set with continuous attribute values, there are two problems need be solved. One is how to estimate the missing data; the other is how to deal with the continuous attribute values. In this paper, we proposed a variable precision fuzzy rough set model with the similarity coefficient α and the accuracy threshold β. It used the fuzzy incomplete upper and lower approximations to estimate the missing data. It considered the impact of the noise data by the tolerance relation in variable precision theory. Iris data set in UCI database was taken as an example to show the effectiveness of the proposed algorithm.