Abstract: The present information era needs knowledge discovery from the vast volume of data. As computer technology has developed to greater height, specifically the Internet led to bang of data. Data availability has gone beyond the human capability of absorption. This increase in enormous volume and varied data paves the way for advances in method to recognize, develop and summarize the data. The data set taken from the microarray experiments often contain some missing values which may primarily occur due to scratches or spots on the slide, dust, inadequate resolution, image corruption and hybridization failures. In this paper, a novel approach is proposed for estimating (predicting) missing values using k-NN regression imputation method to handle incomplete data set. The proposed work provides considerably better results when compared to existing work.

Keywords: k-NN with regression, missing value, data mining, microarray.