Errors in Forecasting

Hi Experts
Errors in forecasting helps us for decent accuracy.
We have 6 errors which are MAD,MAPE, MPE, MSE, RMSE & ET.
Can you explain importance of each error in forecasting?
I found MAD and MAPE are used significantly to judge forecast.If we take multiple products in the selection ID which error is use to measure accuracy.What is the significance of MSE,RMSE,MPE.
I tried statistical forecast in DP, the historical data is very random and dynamic,i have taken Historical data for 36 months and generated forecast for 12 months.
The output of the forecast is constant even i tried with different Alpha, Beta & Gama Values.But no use its showing constant forecast.I used Auto Model selection 1 and 2.
But as far as my understanding, system should give same pattern as historical data but in our case system is giving constant forecast.
Please advice.
Regards
Vicky

Hello
Your question is about Forecast Accuracy Measurements
I give you some information from SAP SCM DP Help about this moment:
Univariate forecasting allows you to measure the forecast error in six ways:
Mean absolute deviation (MAD)
Error total (ET)
Mean absolute percentage error (MAPE)
Mean square error (MSE)
Square root of the mean squared error (RMSE)
Mean percentage error (MPE)
The system calculates the forecast errors by comparing the differences between the actual values and the ex-post values.
Use the forecast error measurements to help you evaluate the accuracy of the forecast.
You select them in the univariate forecast profile. To set upper limits for the forecast error measurements, use the Diagnostic group in the univariate forecast profile; when a limit has been exceeded, the system issues an alert.
You can also select the forecast error measurements on the demand planning desktop, under the Errors tab of the Forecast view.
You can find defenitions and formulas for all this errors in SAP SCM APO DP Help. Search it by term Forecast Accuracy Measurements.
Hope it will help you.
Best regards, Andrey.

Similar Messages

  • ¿Should I go for less desviation media absolute or for less total error in forecast?

    Hello,
    I have started using forecast process for automatic reorder level, I have set the system to use automatic mode and I used also "parameter optimization" in order to get the less DMA because it is an indicator for an accurate forecast, but since total error is other factor to take into account, some materials after forecast process without "parameter optimization" get less total error. I am a little confused which route should I take. I guess I am missing a concept.
    Thanks
    Jorge G

    JJMack wrote:
    ECC Ram that cheep back in the days you remember the cost. If you got the cash the 64bit OS can utilize the RAM these days.
    Yes, more is better, and much more is much better.
    If the system were struggling to get through anything I throw at it, the decision would be easier.  But it's not, really.
    Of course, I could toss out the pocket-change $35 now and revisit the decision in the future...  Maybe prices will break even more.
    -Noel

  • Error in Forecasting

    Hi Guys
    We are facing problem while carrying out forecasting in production system . Forecasting is giving dump .
    These are the details for analysis done . Any pointer regarding the resolution will be highly appriciated , we need to get this resolved asap.
    We find that there is table /SAPAPO/TSAREATE that has the entries for each Planning Area
    MANDT PAREAID                         DOCTAB                                      SELTAB                                      S                                                                               
    600   PA_PROD                                   /1APO/F600000121               /1APO/S600000122      
    We find that the Planning Areas PA_PROD & PA_QUALITY  have the same DOCTAB & SELTAB tables in SCP whereas its two separate tables for each Planning Areas in SCQ. The forecasting is working fine in SCQ. We feel that this is the issue causing the problem.
    The reason being that the Forecasting Program is using a program that needs inputs for the field SEL_9ABOMIO which is not available in /1APO/S600000122 but is available in the corresponding tables in SCQ. The dumps mention this field being missing and resulting in the dump.
    These were some of the analysis when we tried to debug the forecasting program .
    If , any one has faced similar issues , please inform about the resolution .
    Regards
    bhupendra

    Hi,
    Then i Guess your PA is not setup properly especially with the forecasting settings. Check your planning area and run the consistency checks. Also pls check if you have carried out all the steps in master profile creation and assignment of selection ids to forecast profiles.
    Thanks
    Mani Suresh

  • Forecast function in integrated planning

    In forecast function of integrated planning we need to fix forecast period time interval in tcode(RSRHIERARCHYVIRT). otherwise it will show error like
    "Forecast period doesnot fall with in F4 time interval" if we fix that in respective tcode mentioned above. then it will error free.
    But my doubt is we need to fix past period time interval also in the same tcode. because we need both forecast period and past period in forecast fucntion.
    How we can fix both at a time, i need to fix error urgent. pls help me
    I will assign very good points

    Hi,
    Please visit the link below
    link : [http://help.sap.com/saphelp_nw70/helpdata/EN/a4/42e03fc2269615e10000000a155106/frameset.htm]
    Please update the exact requirement.
    Thanks
    Pratyush

  • APO Forecast report

    I need to calculate std deviation and std error in forecast accuracy in the report. How do I calculate ?
    Please help.
    Thanks
    Shalini

    Hi,
    You can write routines in Update Rules, the formula related thinhs you discuss with your Functional APO team then write, here you need to consider somay constraints..
    Thanks
    Reddy

  • Forecast -Bacground Job issue

    Hi,
    Working on SCM 4.1
    We are getting one issue regarding the Forecast Run in the Background...
    After the Background Job run --- for Forecast --- we are getting Error message for some of the SKU-Locations...
    Error --->  "Forecast Values will be negative"
    Becasue of the above error the Forecast isnot updateed in the Forecast Key fig !
    If we run the same SKU-Location -- under the same Forecast profiles and settings --->> It is giving the Forecast values
    And in the message Tab --> it is givin the same message ""Forecast Values will be negative"
    But in foregroubd it is giving the forecast
    So in Background job why system isnot populating the Forecast in the planning Book?
    Is there any specific reson? How can we solve this?
    Thx in advance.
    Regards,
    Rajesh patil

    Rajesh,
    It is normal to see the error message "Forecast Values will be negative" due to decreasing trend. In this case, the negative values just dont get stores in the stat forecast KF, instead it will be null.
    Now to the question of why positive values are not written in background. Please make sure you run forecast at the same level both in Interactive Planning and background as well. Lets say if you use Product - customer in the planning book, select only these levels in the forecast batch job as well.
    Let me know if this helped.
    Hari Vomkarey

  • How can I see the MAD results in my planning book

    Hello Gurus:
    I'm setting up some statistical models in APO DP and running some test data through them.  How can I see the results of the MAD to verify if the models are calculating correctly?  I was hoping I can see the monthly MAD values in a key figure in my planning book?
    Does anyone know?
    Andrea

    Hello Andrea,
    As far as I know, I don't think there's a standard key figure or macro could acheive this.
    MAD or other errors in forecast are not saved as time series, so I don't know why you want to see them as time series.
    You can see the forecast errors in forecast result screen, and if you push comparision button, you can see forecast errors for several last runs.
    But I don't know how to put these value into planning book, unless you want to apply your own logic ...
    Best Regards,
    Ada

  • Result Row calculation

    Hi  result rows calculation is not happening properly in result rows as per the formula .
    Formual for Absolute Error =  Final forecast - Actuals
    Follwing is the report out put
    Material                    Period01
    10324401     Final Forecast        733,490.000 LB
         Actuals              555,583.67 LB
         Absolute Error       177,906.33
    10326795     Final Forecast        1,306,894.935 LB
         Actuals               1,406,158.27 LB
         Absolute Error         99,263.33
    10329435     Final Forecast         188,000.000 LB
         Actuals               192,000.00 LB
         Absolute Error        4,000.00
    10580325     Final Forecast        76,105.938 LB
         Actuals               215,999.82 LB
         Absolute Error        139,893.89
    10580427     Final Forecast        158,733.076 LB
         Actuals                0.00 LB
         Absolute Error         158,733.08
    Overall Result Final Forecast        2,463,223.949 LB
         Actuals                2,369,741.76 LB
         Absolute Error         579,796.62
    In the above report , Material wise calculation for Absolute error is correctly calculating ,  BUT when it comes to Over all Result  its just summing up all  materials Absolute error and showing .
    What i want in  over all result  is Absolute error should be calculated as overall Final Forecast - Over all actual ( I.e formula should be applied to over all result also ).
    I tried with exception aggragation and other stuff but nothing is working .
    Can any one Help me  to apply the formula for result rows also ...
    Helpfull inputs will be appriciated with points .
    thanks in advance

    Hi All  thanks for ur replies .
       I solved my issue  as  following  .
    Formual for Absolute Error = Final forecast - Actuals
    For Absolute error  formula  there are some ' exceptional aggregation ' and 'calculate result value as '  .
    I removed these ' exceptional aggregation ' and 'calculate result value as '  and refreshed the report  now  in result rows
    ' Absolute Error'  is calculating   as desired i.e by using result rows  ' final forecast' value & ' actuals' values in the formulae .
    Regards
    Navenas

  • How can i open the terminal.app in my mac book

    when i click on the terminal icon in my utilities folder, nothing happen nothing open up, why?
    Anyone can help?

    Hello Andrea,
    As far as I know, I don't think there's a standard key figure or macro could acheive this.
    MAD or other errors in forecast are not saved as time series, so I don't know why you want to see them as time series.
    You can see the forecast errors in forecast result screen, and if you push comparision button, you can see forecast errors for several last runs.
    But I don't know how to put these value into planning book, unless you want to apply your own logic ...
    Best Regards,
    Ada

  • Demantra - prediction_status = Young

    Hello,
    I'm using Demnatra 7.3.1.4. Data loaded successfully. I'm trying to generate foreceast. I don’t have any errors but forecast for all combinations is null. I've found out that all records in mdp_matrix table have prediction_status = 98
    Implementation Guide says: „98 (Young) — Sales for this combination are too new to be used for prediction.”
    But I’ve loaded successfully sales history for more than 2 years.
    What can be reason of such status?

    Hi,
    As you are saying that the Prediction_status is 1, it simply means that the combination is forecast-able and there cannot be any issue with Do_Fore.
    Please check the following:-
    Do you have INSERT_UNIT procedure run properly?? Engine does not create rows in SALES_DATA table. If there is an issue in creating the rows in future bucket, the engine won’t be able to write the forecast (even though it generates it at the back end).
    You will find the details in the Engine Log. Would be able to help if you upload the engine log.
    Regards,
    Vicky Vishal

  • Boolean in a combo with "no selection label"

    Does anyone know how to display a boolean value in a combobox with a "no selection label" ? If it's a primitive boolean, false is bound to the control by default. If the property is Boolean it causes Property 'foo' not readable on type java.lang.Boolean

    This is my bean:
    public class Report {
             private boolean forecast;
             public boolean isForecast() {
                      return forecast;
             public void setForecast( boolean reportForecast ) {
                      this.forecast = reportForecast;
    }and this is my jsf page:
    <h:selectOneMenu value="#{report.forecast}" required="true" >
             <f:selectItem itemValue="true" itemLabel="yes" />
             <f:selectItem itemValue="false" itemLabel="no" />
    </h:selectOneMenu>As I wrote before, I would like yet another option in the input, a "no selection option" with label "Pick one..". With the primitive boolean, the field gets instantied with false by default. When I try to use Boolean type, I get the following error when entering the page:
    javax.el.PropertyNotFoundException: /editReport.xhtml @50,108 value="#{report.forecast}": Property 'forecast' not readable on type java.lang.BooleanWhen I try to use autoboxing in accessors (just leave the primitive boolean in the signatures) I get
    javax.el.ELException: /editReport.xhtml @50,108 value="#{report.forecast}": Error reading 'forecast' on type pl.alior.creditrating.entity.Report

  • APO-DP Functionalities

    Hello
    I have few queries regarding the funcionality of APO-DP, these are as under:
    1)Are we able to do editing unconstrained demand and creating a revised work-in-process forecast while retaining the original unconstrained submission.
    2)Is there functionality available for identifying significant differences for a given month between forecast cycle. Example: January forecast for activity in the month of June vs. February’s forecast for activity in the same month of June.
    3)Is the functionality allows for the creation of a final consensus forecast that accounts for adjustments and constraints.
    4)Is distribution of forecaasts possible.
    5)Is there any functionality that accounts for a sales channel’s historical consumption.
    6)Is the system functionality able to calculate statistically generated forecast suggestions using various models checked for best fit.
    7)Does the system support demand driven replenishment for various inventory stocking points.
    Inputs for DDR should be items like safety stock, demand variability, supply variability, lead time, and lead variability. 
    8)Is the system able to generate report on forecast error and forecast accuracy.
    Jitender

    Hello Jitender -
    I have few queries regarding the funcionality of APO-DP, these are as under:
    1)Are we able to do editing unconstrained demand and creating a revised work-in-process forecast while retaining the original unconstrained submission.
    -- you can have a copy of the original unconstrained demand by saving the results and also you can revise it.
    2)Is there functionality available for identifying significant differences for a given month between forecast cycle. Example: January forecast for activity in the month of June vs. February’s forecast for activity in the same month of June.
    -- You are talking about forecast accuracy report. there is an inbuilt functionality for this report as in where u can show the previous forecast plan for the previous month depending on the business requirement.
    3)Is the functionality allows for the creation of a final consensus forecast that accounts for adjustments and constraints.
    -- It all depends on how you want to do consensus. Get the specific final forecasted KF from all the planning books from the various department into the consensus planning book.(thats one way of doing it)
    4)Is distribution of forecaasts possible.
    -- yes it is. you have to look at the proportional factors that are available and distribute the forecast results accordingly.
    5)Is there any functionality that accounts for a sales channel’s historical consumption.
    -- did not get this? if you are talking about using historical data for forecast purpose yes you can otherwise how are you going to do forecast?
    6)Is the system functionality able to calculate statistically generated forecast suggestions using various models checked for best fit.
    -- depending on the forecast model you chose, the smoothing factors u fill in, the system calculates it behind the scene for you and generates the number for you to view.but you ought to understand the logic as to what kind of calculation is done and how it helps you.
    7)Does the system support demand driven replenishment for various inventory stocking points.
    -- gurus Answer.
    Inputs for DDR should be items like safety stock, demand variability, supply variability, lead time, and lead variability.
    8)Is the system able to generate report on forecast error and forecast accuracy.
    yes the functionality has to be activated.
    Hope this helps.
    Regards,
    Suresh Garg

  • Safari can't find insert server Server DNS resolution issues

    I'm getting this error a lot.
    Most of mine are to .gov sites that change their DNS resolution fairly frequently
    Here is a list of the types of sites that often come back with server not found errors.
    http://forecast.weather.gov/MapClick.php?CityName=Paonia&state=CO&site=GJT&textF ield1=38.8695&textField2=-107.59
    http://forecast.weather.gov/MapClick.php?lat=38.89317057287496&lon=-107.59323120 117188&site=gjt&smap=1&unit=0&lg=en&FcstType=text
    http://www.crh.noaa.gov/gjt/
    http://squall.sfsu.edu/gif/jetsat_00.gif
    http://science.nasa.gov/science-news/science-at-nasa/2010/11jan_antimatter/
    http://farmingforum.co.uk/forums/index.php
    Further research has shown that the problem is that Safari has changed how it looks up DNS services between revision 5.0.3 and the one in Leopard as the problem never happened in Leopard.
    Clearing the cache does not help, clearing and resetting cookies does not help. Constant reloading of the page will eventually fix it, whenever Safari finally goes to the proper DNS.
    There is clearly a change in how Safari is looking things up on the DNS but no clue hot to change the behavior to get it back to the Leopard version that worked.

    If you use a router, does it have the latest firmware installed? Safari 5 introduced 'DNS Prefetching' (Safari looks up the addresses of links on webpages and can load those pages faster) which could strain some routers or modems producing a 'latency' in page loading. This was fixed in Safari 5.0.2, but you may see a benefit from getting a faster router or modem. You can read more about that here:
    http://support.apple.com/kb/TS3408?viewlocale=en_US

  • Forcasting table name

    hi ,
    i want to know the forcating table name
    Reply soon

    Pls check the following
    /SAPAPO/PROFTEXT               Forecast Profile Texts
    DBPROF                         Forecast Error with MRP Areas
    ERET                           Forecast/Event Texts
    MPRP                           Forecast Profiles
    OBGV                           Forecast Object Groups
    PROF                           Forecast Errors
    PROP                           Forecast parameters
    PROW                           Forecast Values
    S125                           Forecast history for ex. system
    T440Y                          Forecast weighting factor
    T441A                          Forecast profiles
    T441M                          Forecast transaction control
    MPOP_V                         Forecast Parameters
    Edited by: Manohar Raju on Mar 10, 2008 12:23 PM

  • Error logs in job releases forecast from DP to SNP-"Matl does not exist"

    Hi Everyone,
    I am facing error logs in the batch job which releases forecast from DP to SNP
    Forecast release job - Weekly - Through background job
    Forecast category - FA
    Error 1: - Characteristic combination appears on batch job log with an error that
    "GMC does not exist " Even though it exist in R/3 and APO
    All products with error message "Product does not exist" from do exist and have Forecast in Product View
    Error 2: - Deleted KCC keep showing up on the Demand Forecast release job log as errors even though they are invalid KCC and were deleted previously. How do they get removed from error log so they don't get created again
    Material does not exist on the APO Product Master, in R3 there is a delete flag and the MRP type is set to X1 ,it does not exist in the APO Product Master and does not exist in R3.
    Request you to kindly provide any way forward which may be causing error logs of the job
    Thanks in advance,
    Mrigendra Sharma

    Hi Mrigendra,
    Please re-initialize the SNP planning area (program : /SAPAPO/TS_PAREA_INITIALIZE) and then check the consistency by executing the program: /SAPAPO/TS_LCM_CONS_CHECK. If you observe any inconsistency in the log/spool, then re-run the program: /SAPAPO/TS_LCM_CONS_CHECK to observe all green.
    It is recommended to run the consistency program after any master data change. The initialization program should be run at least weekly.
    Try transfering the data again from DP to SNP to see if the error reappear.
    Hope this helps.
    Regards,
    Nawanit

Maybe you are looking for