Imputation Using Machine Learning Techniques

In today's environment to find the missing value in the data set has become the biggest challenge for the industry people, scientists, academicians and for the researchers. With the incomplete dataset we are not able to apply algorithm to find the result. If the dataset is not complete then only we can apply algorithms and then result can be analyzed to get the efficient output. In this paper we have analyzed some of the missing value generation techniques and algorithms and also the result is compared with among techniques and algorithms which gives the best solution for imputing the data in a incomplete dataset. We have considered some of the missing value handling techniques such as deletion, imputation, EM algorithm, KNN algorithm and Random Forest. Also applied on the incomplete dataset MRI of the brain available in the Research Center of Alzheimer's disease in Washington University.

I. Introduction

Now a day in the digital world nothing can be possible without the data, huge amount of data are in usage for a second. If the data is a complete data then it can be processed and will give the expected result, if it is incomplete then it is quite difficult to get the expected result[1][2]. The incomplete data is considered as the missing data or unknown value. So the researchers, academicians, industry persons, and the medical field persons are focusing to avoid the incomplete data. In many cases there is the chance for having an incomplete data. If there is an incomplete data it is to be converted as a completed data and then the processing is to be done. Many research people have discussed various methods for handling the missing data in the dataset [3]–[6]