84  Imputation

If we are thinking of outliers in the predictor scenario, then we could treat the outliers as missing values and use imputation methods from our missing value section. If we believe an outlier value is a measurement outlier, then we have assumed that the value doesn’t contain any valuable information, Treating it as a missing value that is then imputed.

As we have mentioned earlier, this should be avoided at all costs by trying to fix the data retrieval process that caused the outliers instead. Nevertheless, this is a method that is being used.