NULL value replacement is offered as a transformation function. A literal value, the mean, median, mode or an imputed value joined from another table can be used as the replacement value. The median value can be requested with or without averaging of two middle values when there is an even number of values. The replacement value can also be the analytic data set’s target date value. Literal value replacement is supported for numeric, character and date data types. Mean value replacement is supported for columns of numeric type or date type, with special coding required for date type. Median without averaging, mode and imputed value replacement are valid for any supported type, with distinct SQL generated for computing the median value of numeric, date and other type columns. Median with averaging is however supported only for numeric and date type columns.