Forecasting Accuracy Measurement
- From: "Steve Fletcher" <ste.fletch@xxxxxxxxx>
- Date: 4 Sep 2006 03:31:42 -0700
Hello All,
I'm trying to assess the accuracy of a forecast model that we have used
at work, and I'm unsure how to do it. Rather than give the details of
the scenario I will give a scenario that illustrates a similar
situation.
Imagine I'm a teacher, and I have predicted the percentage scores of my
pupils before they sit a test. I provide a score between 0 and 100
according to my perception of their ability. The pupils then sit the
test and the test is marked independently, and I am given their test
scores. I now have a pair of scores for each pupil, my prediction and
their actual.
How to I measure the accuracy of my predictions? I know I can work out
the mean absolute error, but how do I use this statistic, if at all?
I'm really not bothered how far away from the actual my predition was,
I'm only interested if it was accurate for that pupil. Does this have
some implication over the test I choose? Is it possible for me to
workout the percentage of my accuracy?
Any help would be appreicated.
Many Thanks,
.
- Follow-Ups:
- Re: Forecasting Accuracy Measurement
- From: john2
- Re: Forecasting Accuracy Measurement
- Prev by Date: Re: Multinomial approximation to Poisson ??
- Next by Date: Re: Reef Fish Bob is WRONG: Court Trials ARE NOT hypotheses tests
- Previous by thread: Reef Fish Bob is WRONG: Court Trials ARE NOT hypotheses tests
- Next by thread: Re: Forecasting Accuracy Measurement
- Index(es):
Relevant Pages
|
