Missing Data Processing Based on Deep Neural Network Enhanced by K-Means

Abstract

This paper proposes a neural network model based on K-means to process the problem of data missing. The method first clusters the samples according to the attributes without missing values to get several clusters, and then puts these clusters into different neural networks to predict the missing values. In this paper, the data can be divided into two types: the continuous numerical type and the discrete numerical type. At the same time, corresponding neural network models are established for these two types. We conduct experiments on the dataset called Human Development Index and Its Components, showing our method to be feasible and superior.