Question

Topic: Strategy

Wmape

Posted by vithithuyoanh on 11/25/2014 at 5:31 AM ET 250 Points
Hey everybody!

I am an intern in a company, working for the wMape (Weighted Mean Absolute Percentage Error) process. After all the error calculation I asked myself, how could I calculate the cost, if I could reduce 1% in my wmape?

Does anybody have an idea?

I would appreciate your help! Thank you.
To continue reading this question and the solution, sign up ... it's free!
SUBSCRIBE
Or, SIGN IN

RESPONSES

  • Posted by mgoodman on 11/25/2014 at 12:15 PM Accepted
    Huh? I don't have a clue what you're asking. What is the wMape? And who is the target audience?

    Are you sure this is a marketing issue?
  • Posted by Gary Bloomer on 11/25/2014 at 5:48 PM Accepted
    After a little digging online, and if I understand this correctly:

    1. Isn't WMAPE a weighting for each independent forecasting error, depending on a specific sales volume of a given product?

    2. Can't WMAPE be calculated as ∑{|(A-F)/A|*100*A}/ ∑A (where A = Actual, F = Forecast)?

    3. Isn't WMAPE therefore the total of absolute errors divided by the total of the actuals? Read more in this white paper https://demandplanning.net/documents/dmdaccuracywebVersions.pdf

    Yes? No?

    I dunno if this helps at all but DO know that my brain is beginning to hurt.
  • Posted by mgoodman on 11/25/2014 at 6:17 PM Moderator
    As an engineer by education/training, I am always drawn to equations and quantitative analysis. But I still don't understand the question ... and I can't even begin to figure out what the answer might be.
  • Posted by Mike Steffes on 11/26/2014 at 8:46 AM Accepted
    Just working with what's here- The equation giving the ratio of error to actual... The result of that could be changed 1% by either reducing the error, or increasing the actual.
    So, like most everybody else is saying, there probably isn't enough information given to reach any answer.
  • Posted by Gary Bloomer on 11/26/2014 at 2:37 PM Member
    Here's a link to the equation above: https://ibf.org/forums/index.cfm?fuseaction=read&forum=19&id=48
  • Posted by vithithuyoanh on 11/27/2014 at 5:07 AM Author
    I am sorry, that I did not go further in my explanation.

    The wMape, which means 'Weighted Mean Absolute Percentage Error' is used to measure forecast errors. I, from demand management and the logistics are working for this process.

    For instance:

    1st You forecasted 3 apples and invoiced 3 bananas, so you have a SKU error
    2nd expample: if you forecasted four carottes and invoiced just three, then it is an volume error

    The calculation for that is: ∑ABS(A-F)/ ∑MAX(A;F) A = Actual, F = Forecast

    In the 2nd case it is: ∑ABS(3-4)/∑MAX(3;4) = 1/4 = 0.25 * 100 = 25 % Error

    I calculated monthly in the Sub-Family level to set a new target for our area to decrease our errors.

    My next step was to calculate the BIAS for all the months, which gave me a better analysis of the market behavior of our products.

    What I want to show them now: how much money can we safe, when I could decrease 1% of the error amount? My team tried to find a solution with logistics, calculating and considering the safety stock time, procuration of raw material, etc.

    But now I am a little bit lost. I hope you guys have any better idea, how to calculate the cost for 1% of forecasting error.

    I hope it is not confusing you all. Thanks for your help!


BACK TO TOP

Post a Comment

Related Discussions

Community Info

Top 25 Experts
(Strategy)