If I have missing values in a dataset, I can't just blindly impute them with mean/median/mode or any other technique. I have to identify what kind of missing values they are, namely:
MCAR (missing completely at random) - No relationship between missing value and any other variable
MNAR (missing not at random) - Relationship present between missing values and other variables and missing data is not random.
MAR (missing at random) - Relationship present between missing values and other variables but missing data is random.
To identify the type of missing data I tried the following method. I plotted the following plot:
Here the feature
BsmtQual has nan values so I plotted the nan values against the dependent feature
SalePrice. 0 means it is not a nan value and 1 means it is a nan value. Clearly there is some relationship between the nan values and the target variable as houses with missing values are sold for less price than house with non missing values. So this missingness is MNAR and I would use a MNAR technique to deal with nan values for this feature.
Is this process right? If not is there anyway I can identify what kind of missing data I have (aside from Little's Test)?