An Approach Based on Feature Selection for Missing Value Imputation

Abstract

© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.Today, with the spread of technologies such as the internet of things and data acquisition from sensors, the data obtained has increased. The size of the data produced by other sources, especially digital platforms, is increasing day by day. This increase in data production enables the development of effective artificial intelligence applications and in-depth analysis. However, in many data collection processes, missing values are included in the data set due to operational problems or different reasons. This situation is expressed as a data quality problem in the literature. It is possible that the analysis to be made on this data will be negatively affected by this situation. Various statistical techniques and machine learning-based techniques exist in the literature for filling missing values. In this study, an approach is put forward that suggests missing values imputation based on the consistency of the sample with missing values with other samples in the data set.