Types of Missing data

Data Science, AI and ML
#1

Before dealing with the missing values, it is necessary to understand the category of missing values. There are 3 major categories of missing values.

Missing Completely at Random(MCAR):
A variable is missing completely at random (MCAR)if the missing values on a given variable (Y) don’t have a relationship with other variables in a given data set or with the variable (Y) itself. In other words, When data is MCAR, there is no relationship between the data missing and any values, and there is no particular reason for the missing values.

Missing at Random(MAR):
Let’s understands the following examples:

Women are less likely to talk about age and weight than men.

Men are less likely to talk about salary and emotions than women.

familiar right?… This sort of missing content indicates missing at random.

MAR occurs when the missingness is not random, but there is a systematic relationship between missing values and other observed data but not the missing data.

Let me explain to you: you are working on a dataset of ABC survey. You will find out that many emotion observations are null. You decide to dig deeper and found most of the emotion observations are null that belongs to men’s observation.

Missing Not at Random(MNAR):
The final and most difficult situation of missingness. MNAR occurs when the missingness is not random, and there is a systematic relationship between missing value, observed value, and missing itself. To make sure, If the missingness is in 2 or more variables holding the same pattern, you can sort the data with one variable and visualize it.