Emergence of Bayesian Network as Data Imputation Technique in Observational Studies

Abstract

With data becoming the new oil of the world market, data is essential to key decision makers and businesses to drive growth in all sections of society. Whether it is economy, computer science or medicine, with the advent of information technology in all fields, handling missing data becomes one of the pivotal tasks to ensure accuracy. One of the roadblocks to data driven decisions is missing data. Missing data affects statistical power of data and adds bias, among other ill effects of their presence. The paper reviews existing mainstream methods of data handling and at the same time looks upon Bayesian Network; particularly in Clinical Trials (Observational Studies). We look at different studies that have used a variety of methods to impute data in different medical dataset, along with their advantages and disadvantages that gives a breeze through about data imputation methods and their overview. Along with how the Bayesian network is a model that is proving to be a promising method that can set a benchmark for other data imputation techniques.