|January 6th, 2015, 01:35 AM||#1|
Joined: Jan 2015
From: hong kong
Measuring predictions, statistically
Sorry if this question might appear to be a bit vague.
I am trying to measure the performance of my model, by comparing the actual shots on target attained by a football team against a predicted amount of shots that my model estimates.
I have been reading up on various methods that might help me, but can't quite figure out where to turn.
Should I be using some sort of log likelihood (logistical regression) method?
Or do I need to find the MSE (mean squared error)?
Any help would be greatly appreciated
|January 5th, 2017, 06:43 PM||#2|
Joined: Oct 2013
From: New York, USA
The "Team Performance vs Expected" could have the absolute value of the difference between the two columns, but that might not be useful. I think models need to be compared to each other rather than evaluating one model by itself. Your model has a correlation (r) = 0.4050020254. A positive correlation means the your model predicts in the correct direction.
|measuring, predictions, statistically|
|Thread||Thread Starter||Forum||Replies||Last Post|
|Linear Graph Future Predictions||hein123||Linear Algebra||0||June 7th, 2014 01:57 AM|
|Measuring Reflections||icor1031||Algebra||8||April 3rd, 2014 04:08 PM|
|Bayesian predictions||ybot||Advanced Statistics||4||July 7th, 2013 04:14 AM|
|How to determine the accuracy of predictions?||TheAndruu||Algebra||7||May 13th, 2010 04:39 PM|
|Making a measuring stick for an oiltank||wahlstrom||Algebra||8||December 28th, 2007 12:45 PM|