Side Note: Outlier Correction

In the MRP monitor, you can specify whether the median or mean value is used as the expected value to calculate the standard deviation and variation coefficient.

You can also determine whether an outlier correction takes place.

For the time series of a material, the selected expected value is then used to calculate the standard deviation. All values of the time series that deviate from the expected value by more than the defined multiple of the standard deviation are corrected to the expected value.

Additional columns in the result list show the upper and lower limits with which the values are compared. A checkbox indicates whether a correction has been carried out and the total number of corrections is shown in the “Quantity of outliers” column.

The graphic for the time series shows the original time series (without outlier correction). The upper and lower limits for values as well as the periods in which outliers were corrected can be seen.

In the figure, the consumption quantity in week 36 exceeds the upper limit and the quantity in week 41 falls short of the lower limit. In week 36, therefore, the value 2500 is corrected downward by 1550 to the median value of 950, while week 41 is corrected upward by 750. This results in an overall correction of 2300.

The corrected time series is used, in turn, to calculate the standard deviation, and from this, the coefficient of variation that is used for the XYZ indicator assignment is determined. Performing an outlier correction makes the time series appear more even so that it can be assigned a “better” XYZ indicator.

The selection as to whether the median or mean value is used as the expected value applies only to the key figures described here. For other key figures (such as those used to calculate the MAD), the mean value is always used.