Machine learning methods for handling missing data

Abstract

This FYP is regarding handling missing data and it will show you how to handle such missing data. But what is missing data? Missing data is basically a missing field or answer that someone either forgot to fill it in or simply chose not to fill it in. This can create some issues regarding the completion and analysis of set datasets. So basically this FYP will show you how to handle that missing data and if this is encountered (and trust me, they are always encountered), the program will be able to handle this missing data by using methods such as: 1) Multiple Imputation (MI) 2) Direct Maximum Likelihood (DML) 3) Replacing observed data with missing at random (MAR) They all work differently, some more accurate than the other but some are faster in achieve results than one another. They all have their advantages and disadvantages but these can be used to solve such missing data. The first two methods have their own formula which will be more explained in depth later on in this assignment, whilst the third one is more of an algorithm. As already mentioned, this FYP will cover the basics and solutions as to how missing data is treated and handled and how it is replaced. It will also give you knowledge of the types of methods that can be used to solve such missing data and these methods will be explained in detail. The aim of this FYP is to reduce the amount of problems researchers have to encounter every day when they are trying to solve missing data by replacing missing data using one of the methods that are displayed above.