Loan Prediction Problem - Multiple imputation for missing values?

missing_values
#1

Hello,

So for this course you guys use mean and median for numeric variables and mode for categorical for the missing values. I know that this is generally ok if you have 3% or less missing values in a given variable, however, a more better method is often multiple imputation if percent of missing is > than 3%. Is multiple imputation a good option for this course?

Also, related to this…I know that when you use something like FCS method, you use regression for your continuous variables but I sometimes see people using discriminant method for ordinal data…is this correct?

0 Likes

#2

Also, a second question, are missing values always handled before dealing with outliers?

0 Likes

#3

Where 3% comes from? any reliable source?

0 Likes