• We are pleased to announce that the winner of our Feedback Prize Draw for the Winter 2024-25 session and winning £150 of gift vouchers is Zhao Liang Tay. Congratulations to Zhao Liang. If you fancy winning £150 worth of gift vouchers (from a major UK store) for the Summer 2025 exam sitting for just a few minutes of your time throughout the session, please see our website at https://www.acted.co.uk/further-info.html?pat=feedback#feedback-prize for more information on how you can make sure your name is included in the draw at the end of the session.
  • Please be advised that the SP1, SP5 and SP7 X1 deadline is the 14th July and not the 17th June as first stated. Please accept out apologies for any confusion caused.

Propensity model

E

entact

Member
I am reviewing the chapter on GLMS and I was wonder what the best approach would be to model something that had 6 outcomes. If you were modelling the decision by a person where the options were fixed to 6 and you had loads of factors that influenced those outcomes.

For example, a poker game where a player can decide whether to fold,call,raise(1 big blind), raise(2 big blinds) and raise all in.

Thanks
 
You are looking for multinomial logit or probit models - I think probit is probably a better choice as the logit model assumes that preference for choice is not influenced by addition or removal of other choices, which seems unlikely in the context of a poker game.

The problem with datasets with large numbers of parameters is that you are very likely to overfit, as there will be some set of parameters that fit the data very well.

An alternative to attempting a regression would be to define a loss model and calculate the loss under each possible choice. However, to look ahead several rounds means the model requires probabilities for choices...
 
Thanks Calum but is a probit model not restricted to just to 2 dependent variables whereas I am looking to model more than 2 outcomes.

Could you advise how the model would be structured?

Thanks

You are looking for multinomial logit or probit models - I think probit is probably a better choice as the logit model assumes that preference for choice is not influenced by addition or removal of other choices, which seems unlikely in the context of a poker game.

The problem with datasets with large numbers of parameters is that you are very likely to overfit, as there will be some set of parameters that fit the data very well.

An alternative to attempting a regression would be to define a loss model and calculate the loss under each possible choice. However, to look ahead several rounds means the model requires probabilities for choices...
 
Hi Calum

I also use R - do you have an idea of the code you'd use for something like this? Wondering if you could state that in the context of how you imagine your data would look, how it's aggregated (or not), what your response variables are etc.
 
Back
Top