Ajusted demand modifed by an odd number

Hi Sabrina,

I discussed this with colleagues in IFS Product & Tech (IFS R&D) and it looks like Demand Planner Data Cleansing is at work here.

Data Cleansing is a feature of the Demand Planning forecast engine that cleans historical demand data as part of the forecasting process.

This helps to prevent unusual spikes or one-off demand events from distorting the statistical forecast.

The standard Data Cleansing method performs the following tasks:

1. Calculates the standard deviation of historical demand.
2. Identifies demand values that are more than three standard deviations from the norm.
3. Replaces those extreme values with the 3 sigma limit before forecast model calculations are performed. 

I hope that this helps!

Richard.

Hi Richard

Thnaks for your answer. But how stop this? In 24R2 i haven’t this problem only in 25R2

Hi Sabrina,

Apologies, I should have mentioned that! 😀

You control it in the Demand Plan Server / Advanced Server Settings

Filter for Setting = ‘CleansingMethod’

CleansingMethod: 0=No Data Cleansing, 1=Cleanse Data (according to the Standard Derivation Method). The Default Setting = 1.

Try setting the CleansingMethod = 0 and Adjusted Demand periods should never be automatically adjusted.

Good luck,

Richard.

Thanks richard. I try!!

Hi Richard,

Thank you for the explanation. Recently, I have also been spending more time analyzing how Demand Planning performs its calculations and how some of the forecasting features work.

I am familiar with the Data Cleansing functionality and understand the concept of identifying outliers based on standard deviation. However, I am curious about the actual inputs used for the standard deviation calculation.

As I understand it, the calculation follows the standard formula:

What I would like to understand better is the value of n in this case. How much historical demand data is included in the calculation? Does the system evaluate the entire demand history, or only a specific period?

Is this period determined by Demand Planning Server settings, forecast horizon settings, or some other configuration parameter? Also, is it possible to influence or modify this behavior through any of the Advanced Settings?

I have reviewed the IFS documentation on Data Cleansing (ifsdoc/documentation/en/DemandPlanning/AboutDataCleansing.htm), but I could not find any details about how the sample size (n) is determined or which historical periods are included in the standard deviation calculation.

I would appreciate any additional details on how IFS determines the dataset used for the standard deviation calculation.

Thanks in advance!
Daniela