Thanks Joe. This is good. But can you please tell me how to set up the
variable as either fixed effect or random effect?
*as per the documentation, rhierBinLogit model is:*
P = exp(Xij BETAi) / (1 + exp(Xij BETAi)
where, BETAi = Z Delta[i,]
for keyword i and week j.
Here all X are modeled as random effects. How do I model some X as random
and some as fixed effect coefficient?
Thank you!
Kiran
2009/3/14 Joseph Retzer <joe_retzer@yahoo.com>
> You may want to take a look at:
>
> rhierBinLogit MCMC Algorithm for Hierarchical Binary Logit
>
> in the bayesm package.
>
> Cheers,
> Joe
>
> --- On *Fri, 3/13/09, Kiran BM <kiranbm@gmail.com>* wrote:
>
> From: Kiran BM <kiranbm@gmail.com>
> Subject: [R] Hierarchical Bayesian Modeling in R
> To: r-help@r-project.org
> Date: Friday, March 13, 2009, 1:47 PM
>
> Hi Friends,
> I'm trying to model the consumer decisions (Click-Through Rate and
> Conversion) in Search Engine Advertising using a hierarchical Bayesian
> binary logit. The input data is the weekly CTRs and Avg. Position for each
> search keyword.
>
> CTR is
> modeled as (for each keyword i and week j):
>
> Pij = exp(C + Bi x Positionij + A1 x Lengthi + A2 x Brandi + A3 x
> ProductSpecifici) / [1 + exp(C + Bi x Positionij + A1 x Lengthi + A2 x
> Brandi + A3 x ProductSpecifici)]
>
> The Position coefficient Bi is in turn allowed to vary along the population
> mean (B1) and the keyword characteristics as:
>
> Bi = B1 + K1 x Lengthi + K2 x Brandi + K3 x ProductSpecifici
>
> How can I model this in R? Which function in R is used to do the
> Hierarchical Bayesian Binary Logit modeling. Please help.
>
> Thank you!
> Kiran
>
> [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help@r-project.org mailing list
> stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
>
>
[[alternative HTML version deleted]]