Zeev....good morning, technically this is OT, but I thought you may want to give it a whirl:)
I have a friend on another stock board who posed this query. I was lost at Chi Square, but it seems like a problem you have the intellectual mastery to answer.....here it is....
I have some "count" data with lots of zeroes - number of people getting on and of busses at each stop. Seems like it would be suited for a poisson regression. The independent variables are things like pop. density around the bus stop and expected level of transit service - not that that really matters, but just fyi.
Now, with a plain ol' vanilla OLS linear regression, the r-squared tells you something very valuable. You can have a good t-statistic for your coefficients, but a really low r-squared, and, imo, such a "model" is worthless for its predictive power when the r-squared is low.
So, here's my question: since a poisson regression has no r-squared, what do you suggest as an indicator of the predictive ability of a model that is based on a poisson regression?
Market Data 
Markets 
AI
