comparing forecast vs actual

Six Sigma – iSixSigma Forums Old Forums General comparing forecast vs actual

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
  • #48437


    What is the best way to statistically analyze forecasted vs actual sales data to determine the accuracy of the forecast tool? I have a about 5 data points (one per quarter) for 3 different product families. Also since the same forecasting tool is being used on all 3 product families am I able to combine the data sets to have a larger sample size? Any help would be appreciated.


    Erik L

    There are three standard measures that can be used to assess the accuracy of your forecasting predictive model against a time series of actual data.  MAPE, MAD, and MSD are the standards that are typically used. 
    MAPE (Mean Absolute % Error) is calculated as:                                                                                                                                                                   (‡” |((Actual-Predicted)/Actual)|/n)*100
    MAD (Mean Absolute Deviation) is calculated as:
    MSD (Mean Squared Deviation) is calculated as:
    MAPE will provide accuracy of the forecasting model on a % basis.  MAD will provide the accuracy on a unit basis, and of the measures, MSD, is most affected by large deltas between the model and the actual data.  Were you looking for something else?  Based off of your other question, regarding the efficiency parameter, Ifm assuming you have already been exposed to these more basic measures.



    Might you not prepare a Shewhart chart of the observed differences between actual and forecast to assess the differences?

Viewing 3 posts - 1 through 3 (of 3 total)

The forum ‘General’ is closed to new topics and replies.