search

LeoEgidi / footBayes

An R package for many football models
38 stars 7 forks source link

extending the framework for additional variables #3

Open fine-lemur opened 11 months ago

fine-lemur commented 11 months ago

Hi -- nice encapsulation of popular models. Easy to work with.

What's the most straightforward way to keep the simplicity, but add in other variables -- for example game week, or externals like weather?

LeoEgidi commented 11 months ago

Thanks!

The most easy way would be to include these covariates directly in the linear predictors for the log-average scores. For the Bayesian models, you could first fit a Stan model with the function 'stan_foot', then retrieve the original Stan code, and manage this code by including these new variables directly in the specification of log \lambda's. Hopefully it is clear!

In the next versions of the package we'll give the possibility to include more and more covariates, similar as the ones you suggest!

Best. Leo

Da: fine-lemur @.> Inviato: venerdì 4 agosto 2023 13:08 A: LeoEgidi/footBayes @.> Cc: Subscribed @.***> Oggetto: [LeoEgidi/footBayes] extending the framework for additional variables (Issue #3)

Hi -- nice encapsulation of popular models. Easy to work with.

What's the most straightforward way to keep the simplicity, but add in other variables -- for example game week, or externals like weather?

— Reply to this email directly, view it on GitHubhttps://github.com/LeoEgidi/footBayes/issues/3, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AED7PLL34B7HWR5IAMMMKCDXTTJ27ANCNFSM6AAAAAA3EBXDPU. You are receiving this because you are subscribed to this thread.Message ID: @.***>

fine-lemur commented 11 months ago

helpful pointers. thanks. hello, learning curve! Co-variates would be top of my list of feature requests -- having a generalised way to explore that would be a very high-value addition. Do you have any ETA on a version with that as a feature? I know you've mentioned other new features, up to and including scraping odds... my vote would be to concentrate on adding generally-useful model-focussed features (such as the co-variates) as a higher priority as there are a million scrapers and bookmakers, and its a widely-practised art -- whereas the bayesian craft is more thinly-distributed ;-)

LeoEgidi commented 10 months ago

No beta versions right now, me and my collaborators are working on it😉

Anyway stay tuned, I am going to submit the new version of the footBayes package, 0.3.0, in these days, with new features and model choices.

Thanks a lot

Da: fine-lemur @.> Inviato: lunedì 7 agosto 2023 14:36 A: LeoEgidi/footBayes @.> Cc: EGIDI LEONARDO @.>; Comment @.> Oggetto: Re: [LeoEgidi/footBayes] extending the framework for additional variables (Issue #3)

helpful pointers. thanks. hello, learning curve! Co-variates would be top of my list of feature requests -- having a generalised way to explore that would be a very high-value addition. Do you have any ETA on a version with that as a feature? I know you've mentioned other new features, up to and including scraping odds... my vote would be to concentrate on adding generally-useful features (such as the co-variates) as a higher priority as there are a million scrapers and bookmakers, and its a widely-practised art -- whereas the bayesian craft is more thinly-distributed ;-)

— Reply to this email directly, view it on GitHubhttps://github.com/LeoEgidi/footBayes/issues/3#issuecomment-1667777362, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AED7PLPXK4E2QCP3PNPNMATXUDOMLANCNFSM6AAAAAA3EBXDPU. You are receiving this because you commented.Message ID: @.***>

fine-lemur commented 10 months ago

awesome. in time (almost) for the new season ;-)