03 - Demand Planning PDF
03 - Demand Planning PDF
03 - Demand Planning PDF
In this case study the full demand planning process will be explained and
organized: from the configuration of the demand planning, choosing of the
adequate forecasting methods and the execution of the demand planning
till its release for the supply network planning.
Authors
Hristina Ivanova
Version
2.31
Last Update
March 2017
© SAP SE
CASE STUDY
Process Overview
During demand planning the historical data (such as billing quantity or volume of sales) gets
adjusted for the future by means of the forecasts. The result of the forecast is the demand plan.
The plan can be released in order to generate demand for the products in the respective locations
and periods (planned independent requirements). The calculated demand can be satisfied by
means of internal/external production. Small alterations in demand planning lead to large changes
in production planning. The goal of the demand planning is to provide the most correct forecasts
for the sales volume.
The graphic below displays the complete process and at which point we are currently
© SAP SE Page 2
CASE STUDY
Table of Contents
Process Overview ....................................................................................................................... 2
Step 1: Administration of Demand Planning............................................................................ 4
Step 2: Executing the Demand Planning ................................................................................ 15
Step 3: Other Univariate Forecast Models.............................................................................. 27
Step 4: Multiple Linear Regressions (MLR) .......................................................................... 38
Step 5: Composite Forecast ..................................................................................................... 45
Step 6: Release the Demand Planning .................................................................................... 50
© SAP SE Page 3
CASE STUDY
Task Create one planning area, one planning book and one forecast profile. Time 15 min
Before we start making the demand planning we have to create one planning
area, one planning book, and one forecast profile.
At first we should create our planning area.
A planning area is a central data structure that is used as a storage for
planning data for demand and supply network planning. Characteristics and
key figures with their functions are determined here.
A key figure is a numerical figure that denotes either an amount or a value.
Characteristics are the objects to which operational data is aggregated,
disaggregated, and evaluated.
A planning area summarizes main parameters for the coverage of the
planning. In this way it determines where and how the planning results will
be saved. The demand planning plans only in time series that are saved in the
version of the planning areas. Before you can work with a planning area you
have to create time series objects. This process is also named initialization of
the planning area. The system creates a network that consists of
characteristics and key figures in the liveCache.
To create the time series objects, please follow the menu path:
Advanced Planning and Optimization ► Demand Planning ► Menu path
Environment ► Administration of Demand Planning and Supply
Network Planning
Execute a right click on the planning area 9ADP01 and choose 9ADP01
from the list.
© SAP SE Page 4
CASE STUDY
GBI-##
As a planning version type in GBI-##.
One year ago
The Start Date should be one year ago and the End Date should be one year One year in the future
in the future.
Click on .
© SAP SE Page 5
CASE STUDY
The key figures define the master data, such as products, locations, etc. and
the characteristics define the transaction data, such as forecasts, customer
orders, etc.
GBI_DP##
Type in GBI_DP## for planning book and click on .
Planning Book
In the next screen enter for Planning book text Planning Book GBI_DP## GBI_DP##
9ADP01
and for Planning Area 9ADP01.
Univariate forecast
Moreover enable the checkboxes for Univariate forecast, Multiple linear Multiple linear
regression, and Composite forecast. The forecasting procedures will be regression
Composite forecast
explained later.
Click on in the lower right corner to get to the next step.
© SAP SE Page 6
CASE STUDY
Total Forecast
Click on the Total Forecast and then click on to add the selected key
figures. The key figure appears on the left side and belongs now to your
planning book.
Click on in the lower right corner to get to the next step.
© SAP SE Page 7
CASE STUDY
Mark the corresponding APO Location and APO Product and then click on
to add the selected key figures.
© SAP SE Page 8
CASE STUDY
GBI_DP##
The data view should be named GBI_DP## and described as Data View Data View GBI_DP##
9A52WEEKS
GBI_DP##. Assign to both TB Profile IDs 9A52WEEKS (52 WEEKS) by
means of input assistance. Then press Enter.
Visible
Now you can mark the fields Visible and Input and assign both to the first Input
The first entry you can
entry you can select via the selection feature. In the Table 1 type in select
Demand Planning GBI_DP##. Demand Planning
GBI_DP##
Click on in the lower right corner to get to the next step.
© SAP SE Page 9
CASE STUDY
Total Forecast
Mark the key figure Total Forecast then click on to add the selected key
figures.
Then click on .
© SAP SE Page 10
CASE STUDY
Answer if you are asked whether the planning book and the data
view should be completed. Then click on to go back to the main menu.
Periods: 52 52
History Horizon To: Sunday before the forecast start Sunday before the
forecast start
Periods: 52 52
All in all, the past and the forecast we use cover one year.
Enable the checkbox for Univariate Forecast and enter as a profile Univariate Forecast
GBI_DP##_UNI
GBI_DP##_UNI.
Choose the tab Univariate Profile.
© SAP SE Page 11
CASE STUDY
GBI-##
Version: GBI-##
31 (Season according to
Forecast strategy: 31 (season according to Winters) Winters)
© SAP SE Page 12
CASE STUDY
With the help of the forecast strategy you set the forecasting procedures that
31 (Seasonal Winters´)
need to be applied. Here choose 31 Seasonal Model Based on Winters'
Method. A forecast procedure should be applied which considers seasonal
fluctuations in the market development.
© SAP SE Page 13
CASE STUDY
© SAP SE Page 14
CASE STUDY
The persons who take part in the demand planning process need a special
tool in order to be able to analyze, create, and edit the marketing program.
With the help of the planning books and data views the users are able to
interact with the SAP SCM. The planning table is a perfect tool for these
purposes. It consists of two parts.
On the left there are 4 sections:
• Selection window – for choosing the objects (locations, products,
resources, etc.)
• Selection profile – for saving the chosen objects into your profile
• Planning book – for choosing the planning book and data views
• Macros – we can create a macro for every planning that can, for
example, generate alerts.
© SAP SE Page 15
CASE STUDY
A planning area can have one or several planning books. Every book stands
for one particular role in the demand planning process. Every role can
require other information or a particular display format of the information. It
can also happen that particular roles are not allowed to access some features.
© SAP SE Page 16
CASE STUDY
Every planning book can contain an unlimited amount of data views. Data
views determine the layout of the planning table and are the most important
tool for creating, analyzing, and modifying of the planning data. Every data
view stands for one task that has to be done by one corresponding person
(role).
SAP has prepared many data views for standard plannings.
Now we can start the demand planning. To demonstrate various possibilities
of the forecast strategies, follow the severely simplified examples.
If you are asked, whether a planning version should be installed, press no.
Search for the data view of your planning book in the section Planning
book/Data view, open the navigation tree and execute a double click.
© SAP SE Page 17
CASE STUDY
Now press and type in the following data for the filter:
Show: APO product APO product
DXTR1000 / DXTR2000
APO Product: DXTR1000 and DXTR2000 (via multiple
selection and saving the new window via )
© SAP SE Page 18
CASE STUDY
Click on to save the selection. In this way you can access both products
later without creating a filter one more time.
© SAP SE Page 19
CASE STUDY
Now you see both products at the top left in an area that is called Shuffler.
The next step is going to be assigning the selection to our profile. Click on
to maintain selection profile.
The window Maintain Selection Profile for Users opens. Search for the just
GBI_DP##
configured profile GBI_DP##. It will be on the right. Pull it via Drag&Drop
to the left in your user folder and click on to adopt.
© SAP SE Page 20
CASE STUDY
Now we just have to assign our forecast profile. In the menu press
Settings ► Forecast Profile. Menu bar
GBI_DP##
Choose your Master forecast profile GBI_DP## and click on to adopt.
© SAP SE Page 21
CASE STUDY
A double click on DXTR1000 in the Shuffler. It will load the product and we
can now type in our past data.
© SAP SE Page 22
CASE STUDY
The demand volume is given in the following table. Scroll to the left in the
planning area and start filling out the table with the figures given below. You
should end in your own week.
1 2 3 4 5 6 7 8 9 10 11 12 13
2 6 11 15 19 24 34 24 19 15 11 6 2
14 15 16 17 18 19 20 21 22 23 24 25 26
2 6 11 15 19 24 34 24 19 15 11 6 2
27 28 29 30 31 32 33 34 35 36 37 38 39
2 6 11 15 19 24 34 24 19 60 11 6 2
40 41 42 43 44 45 46 47 48 49 50 51 52
2 6 11 15 19 24 34 24 19 15 11 6 2
*the top line is for guidance only.
The demand volume is cyclic and repeats as you can see every 13 weeks
(except the 36th week). It is so intended. It represents a seriously simplified
example for a seasonal fluctuating demand volume.
The 36th week exemplifies an outlier. This is a past value that lies outside
the awaited range of values. Outliers can falsify the forecast results. By
means of the outlier control feature this value will be corrected. We have
appointed for this purpose the Ex-post-Method before. Using this method the
system utilizes the Ex-post-Forecast to determine the tolerance range. If a
past value lies outside the tolerance range, the system understands it as an
outlier and corrects it.
The width of the tolerance range for the automatic runaway value control is
defined through the sigma factor. It sets the amount of the tolerated standard
deviations. The smaller the sigma factor, the lesser the tolerance and the
bigger the amount of the measured and corrected runaway value. The
standard sigma factor is 1,25. If you set your own sigma factor, SAP
suggests the setting between 0,6 and 2.
In the same way like for DXTR1000 produce the demand forecasts for
DXTR2000.
Double click on DXTR2000. If you are asked, whether you would like to
save, click always on .
© SAP SE Page 23
CASE STUDY
1 2 3 4 5 6 7 8 9 10 11 12 13
1 1 1 1 3 3 3 3 5 5 5 5 5
14 15 16 17 18 19 20 21 22 23 24 25 26
7 7 7 7 9 9 9 9 11 11 11 11 11
27 28 29 30 31 32 33 34 35 36 37 38 39
13 13 13 13 15 15 15 15 17 17 17 17 17
40 41 42 43 44 45 46 47 48 49 50 51 52
19 19 19 19 21 21 21 21 23 23 23 23 23
*the top line is for guidance only.
As you can see the demand rises with time. It is a simple trend example.
Double click again on the selection DXTR1000. If you are asked, whether you
would like to save, click on . Then click on for Univariate
Forecast.
If you asked again about a forecast profile, choose your own (GBI_DP##) GBI_DP##
© SAP SE Page 24
CASE STUDY
You can also observe the results in the form of a diagram if you click on
.
© SAP SE Page 25
CASE STUDY
It is obvious that a forecast has been conducted. 52 weeks in the future are
sufficient. The seasonality was taken into consideration. As expected, the
values of the forecast should correspond exactly to our past data. Click on
to save your results and then on to go back to the main menu.
© SAP SE Page 26
CASE STUDY
Choose your planning area 9ADP01 and your Master Profile GBI_DP##. In 9ADP01 / GBI_DP##
50 (Forecast with
the tab change to Univariate Profile and change the forecast strategy to 50 automatic selection)
(Forecast with Automatic Model Selection) and the smoothing factors
0,3
Alpha and Beta to 0,3.
© SAP SE Page 27
CASE STUDY
© SAP SE Page 28
CASE STUDY
The system has different statistical tests and test combinations that can
determine a model. To choose the best forecast profile, the system analyzes
the past data. If no explicit time series processes are detected, the system will
choose the continual model automatically. In our case the system detects the
seasonal changes and selects the seasonal strategy.
You don´t need to save this forecast.
In the tab change to Univariate Profile and change the forecast strategy to 21 21 (Holt´s Method)
(Holt's method).
© SAP SE Page 29
CASE STUDY
If you asked again about a forecast profile, choose your own (GBI_DP##) GBI_DP##
The forecast is conducted according to the Holt’s method and is suitable if the
past values are depicted through a rising or falling trend. Click on
for graphical view.
© SAP SE Page 30
CASE STUDY
As you can see this time the past data was rising. With the given pattern the
trend procedure predicts rising of the future demand data.
We have given Alpha a small value because we wanted to involve the past
data into the forecast in the most balanced way.
The smoothing factor balances the past data in a way that the most up-to-date
values are considered stronger than the older ones.
You don´t need to save this forecast.
How fast the forecast reacts to changes depends on the smoothing factor. If
Alpha is 0, the new average value equals the previous one. In that case the
basic value that was once calculated does not change. It means that the
forecast does not react to the current consumption data. If Alpha is 1, the new
average value will equal the last consumption value.
Reasonable values for Alpha are between 0.1 and 0.5.
Now change the forecast strategy to 94 (Linear Regression) and execute the 94 (Linear Regression)
forecast with DXTR1000 in the demand planning.
© SAP SE Page 31
CASE STUDY
© SAP SE Page 32
CASE STUDY
Choose DXTR1000 again and delete the whole available data. Scroll back to
the week again and type in the following data:
© SAP SE Page 33
CASE STUDY
Change the forecast strategy to 41 (Holt and Winters' Exponential 41 (Holt and
Winters´Exponential
Smoothing) and change the smoothing factors (Alpha, Beta, Gamma) again Smoothing)
to 0,1 each. 0,1
© SAP SE Page 34
CASE STUDY
© SAP SE Page 35
CASE STUDY
© SAP SE Page 36
CASE STUDY
© SAP SE Page 37
CASE STUDY
In the tab choose MLR Profile. Type in GBI_DP##_MLR for the Profile and GBI_DP##_MLR
MLR PROFILE
describe it as MLR Profile GBI_DP##. GBI_DP##
9ATOTFC
In the section past for the dependent key value Y choose as Key Figure
9ATOTFC (Total Forecast) and as Version enter GBI-##. GBI##
© SAP SE Page 38
CASE STUDY
Price
Promotion
In the popup enable checkboxes for Price and Promotion and click on to
continue.
Look at both time series in the popup Time Series Maintenance and click on
to adopt.
© SAP SE Page 39
CASE STUDY
Next Monday
In both cases look for the next Monday of 11 months ago and choose the - 11 months
+ 4 weeks
date 4 weeks later as Start.
© SAP SE Page 40
CASE STUDY
Multiple Linear
In the tab Master Profile enable the checkbox for Multiple Linear Regression
GBI_DP##_MLR
Regression and choose the just now created profile GBI_DP##_MLR, if
this hasn´t been done automatically.
Click on to save the master profile and then on to go back to the main
menu.
© SAP SE Page 41
CASE STUDY
Select the product DXTR1000 and overwrite its data using the following
values:
1 2 3 4 5 6 7 8 9 10 11 12 13
16 23 22 23 23 25 25 25 24 23 22 23 23
14 15 16 17 18 19 20 21 22 23 24 25 26
33 35 35 35 34 34 34 34 34 35 35 35 35
27 28 29 30 31 32 33 34 35 36 37 38 39
44 45 45 45 46 45 45 45 45 47 46 47 47
40 41 42 43 44 45 46 47 48 49 50 51 52
37 34 34 34 34 35 35 35 35 34 34 34 34
Click on for Multiple Linear Regression (It may happen that here you will
be asked to select your forecasting profile one more time. You may also be
asked whether you want to save your changes, click on .
© SAP SE Page 42
CASE STUDY
If you click on to switch Parameter On/Off in the table view, you can see
the fit measures. (If your current year has 53 weeks, your results might be a
little different)
© SAP SE Page 43
CASE STUDY
© SAP SE Page 44
CASE STUDY
This feature unites the forecasts that have different forecasting procedures
(for example, time series, multiple linear regression, and/or estimation
procedure) and were made for particular brands, product lines, or single
products. Every forecast exploits the same past data, however they have
different procedures. The goal is to combine single forecasts with different
strengths to obtain one forecast. You can either find the average value of the
forecasts by means of equal emphasis of the single forecasts, you can assess
every forecast differently, or you can change the weighting of the single
forecasts through time. Combination of the forecasts is supposed to provide a
better forecast for a company. It became evident that the composite forecasts
that are based on different mathematical and/or estimation procedures are
often superior to the single forecasts and their underlying procedures.
Click on .
Select GBI_DP##_UNI from the list. You can select several profiles here if GBI_DP##_UNI
© SAP SE Page 45
CASE STUDY
Click on .
Now we can create another weighting. In our example we have chosen 40 for Univariate: 40
MLR: 60
the Univariate and 60 for the MLR.
© SAP SE Page 46
CASE STUDY
Enable checkbox for Composite Forecast and choose the just now created Composite Forecast
profile GBI_DP##_COM, if this hasn´t been done automatically. GBI_DP##_COM
Click on to save the master profile and then on to go back to the main
menu.
Select the product DXTR1000 and start the Composite Forecast by clicking
on . It might happen that you will be asked again to select your forecast
profile.
© SAP SE Page 47
CASE STUDY
Now the selected forecast profiles will be shown and the results will be
displayed in the table.
In our example the results will look like this:
© SAP SE Page 48
CASE STUDY
© SAP SE Page 49
CASE STUDY
Task Release the Demand Planning for the Supply Network Planning. Time 15 min
Before releasing the demand volume we shall adjust the forecast for the last
time.
© SAP SE Page 50
CASE STUDY
For DXTR1000:
For DXTR2000:
© SAP SE Page 51
CASE STUDY
For that we need a Forecast Split and Distribution Profile, so the demand will
be well distributed within the periods.
GBI-##
For the distribution function type in GBI-## and click on .
Our demand volumes are planned for every week. We want that they will be
distributed to the first (Monday) out of 5 working days (5 periods) (holidays
will be skipped automatically).
For description enter Distribution Function GBI-## and for length enter 5. Distribution Function
GBI-##
Click on . 5
© SAP SE Page 52
CASE STUDY
Here you can distribute your demand to 5 periods you set up before. For the
1
first period type in 1.
Save the data by clicking on .
© SAP SE Page 53
CASE STUDY
GBI-##
For profile enter GBI-## and for description enter Period Split Profile GBI-
Period Split Profile
##. Click on to add split level. GBI-##
52
In the created row enter 52 for the number of periods. As mentioned before we
Week
want to distribute weeks to days. That is why we select week as starting
periodicity and day as target periodicity. For the Distribution Function enter Day
GBI-##
the already made Distribution Profile GBI-##.
Click on to save and then on to simulate period spilt profile.
© SAP SE Page 54
CASE STUDY
Current date
In the following screen select current date as starting time and the date in a Today in a year
GBI-LOC
year from today as end time. For the calendar you should take GBI-LOC and
for the quantity per interval enter 100. Now execute the simulation. 100
Click on to execute.
© SAP SE Page 55
CASE STUDY
© SAP SE Page 56
CASE STUDY
GBI-##
Planning Version: GBI-##
9ATOTFC
Key Figure: 9ATOTFC (Total forecast)
Target:
GBI-##
Planning Version: GBI-##
FA
Category: FA (Planned Independent Requirements)
Horizon:
From date: next Monday next Monday
Object Selectn:
DXTR1000
Product: DXTR1000 and DXTR2000 (through DXTR2000
Multiple selection) DL10
Location: DL10
Results Log
Enable the checkbox for Results Log.
Click on to execute
© SAP SE Page 57
CASE STUDY
If you get this notification and the to date changes, click again on .
© SAP SE Page 58
CASE STUDY
Now you can check the results. Click on to go back to the menu and go to
the product view:
Menu path
Advanced Planning and Optimization ► Production Planning ►
Interactive Production Planning ► Product View
In the popup “Welcome to the Product View” choose SAP_SCO and confirm it
by clicking on .
GBI-##
For the Planning Version enter GBI-##, for the product DXTR1000, and for DXTR1000
DL10
location DL10. Click on to continue.
© SAP SE Page 59
CASE STUDY
Now you can control the demand planning for DXTR2000 as well.
© SAP SE Page 60
CASE STUDY
If you spotted an error, you can delete the just made transaction data and
execute the planning one more time. Use the transaction:
Menu path
Advanced Planning and Optimization ► Supply Network Planning ►
Environment ► Delete Transaction Data
Here in Objects Selection choose By Categories und type in FA. For the
Planning Version type in GBI-##, then the Product Numbers DXTR1000 and
DXTR2000 (via multiple selection), and for the Location type in DL10. Press
Execute. Check the selected data in the log and choose Delete selected objects.
© SAP SE Page 61