Missing Data Imputation Using Ensemble Learning Technique: A Review

Abstract

For the past two decades, several studies have been conducted on missing value imputation in bioinformatics and offered the best method or approach for handling the datasets with missing values. When the datasets have a lesser amount of missing attribute values in the entire database, the missing attribute values be able to remove from the dataset without taking a noteworthy influence on the final mine. However, if a huge number of attribute values are missing, suspicious attention should be given to handle these kinds missing data because the entire dataset will lose their valuable information and the quality of the datasets. In particular, datasets have more than one missing attribute value disturb the algorithms performance. Missing value imputation method’s aim is to provide high-quality dataset without loss of any valuable information intelligently where the missing values are smaller or larger. Meanwhile ensemble learning techniques are achieving high performance in data mining task for the past few years. Researchers, therefore, prefer to work on the imputation of missing data using ensemble learning, a technique that cannot be ignored nowadays because missing data in bioinformatics datasets are rapidly increasing. Ensemble learning aim is transforms from weak learner to strong learner. Those ensemble techniques can process a massive amount of data in an efficient manner. This paper concentrates on the review of missing value imputation techniques and ensemble learning models for analyzing biological data.