iSixSigma

Predicted vs. Actual for Problem Tickets

Six Sigma – iSixSigma Forums Old Forums General Predicted vs. Actual for Problem Tickets

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #43652

    boettler
    Member

    I have created a model that attempts to predict the number of problem tickets resolved in a month.  A month is defined as 21 working days.  I wanted to check the predictability of model at several points (5,10, 15, and 20 days).  Does anyone know what is the best way to analyze this data to determine if the prediction is statisitically different than the actual value at day 21?  I do not think I should use the ANOVA test in this case.
    Thanks for your help.

    0
    #138745

    Robert Butler
    Participant

      When you say you have a model are we to assume it is a regression model?  If so, every regression model has an estimate of root mean square error associated with it.  A rough estimate of prediction error (+-95%) is 2 times this term. 
      Run your equations, generate the prediction, take the prediction +- this term and you will have a rough estimate of the error associated with the prediction.  Let your process run and see what you have at 5,10,15 and 20 days.  If what you get is inside these prediction limits then it is reasonable to say that your model works and that there is no difference between observed and predicted.

    0
    #138749

    boettler
    Member

    Thanks you.  I guess it time to take out the management science materials again! 
    Rob

    0
Viewing 3 posts - 1 through 3 (of 3 total)

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