The post compilation enables you to handle missing values after compilation. In earlier releases of the software, these had to be performed manually after compilation, resulting in two branched datasets which was never used. Now you can compile, set a threshold for allowed missing values and a method for imputing the them in one go. If you check the Keep Missing Indices it is still possible to change the imputation method (Raw Data | Replace Missing Values) or allow fewer missing values (Data Set | Filter Data Set ) later.