R help - Jun 2013 - Package 'ChoiceModelR' - Xbetas.csv? Rbetas.csv? betadraws?

Hope to get some clarity from the creators of a (wonderful) package
ChoiceModelR. I am using the main function 'choicemodelr'.

Everything works.

?choicemodelr says "average of MCMC draws of unit-level model coefficients
are written to Xbetas.csv"
Question 1: Is this a typo? The real file created in the working directory
seems to be "Rbetas.csv"
Questions 2: Are these the averages of the 'use' draws (specified in my
choicemodelr statement)?
Question 3: If I had constraints, I assume these averages will be "taking
into account constraints", right?

?choicemodelr also says that choicemodelr produces an object that contains,
among other things, two elemens:



betadraw

An ni by natt by floor(use/keep) array of MCMC random draws of unit-level
multinomial logit model parameter estimates.

betadraw.c

An ni by natt by floor(use/keep) array of constrained MCMC random draws of
unit-level multinomial logit model parameter estimates.



If I do:

apply(out$betadraw.c,c(1,2),mean)

then what is the difference between the result and the file written to Disk?

One difference I noticed:

apply(out$betadraw.c, c(1,2),mean) contains as many columns as parameters,
not as many columns as the number of levels.

What else?



Thank you very much!


-- 
Dimitri Liakhovitski

	[[alternative HTML version deleted]]

On 07.06.2013 21:25, Dimitri Liakhovitski wrote:
> Hope to get some clarity from the creators of a (wonderful) package
> ChoiceModelR. I am using the main function 'choicemodelr'.
Since we do not know if these "creators" are reading the list, why not
ask them directly?

Best,
Uwe Ligges

>
> Everything works.
>
> ?choicemodelr says "average of MCMC draws of unit-level model
coefficients
> are written to Xbetas.csv"
> Question 1: Is this a typo? The real file created in the working directory
> seems to be "Rbetas.csv"
> Questions 2: Are these the averages of the 'use' draws (specified
in my
> choicemodelr statement)?
> Question 3: If I had constraints, I assume these averages will be
"taking
> into account constraints", right?
>
> ?choicemodelr also says that choicemodelr produces an object that contains,
> among other things, two elemens:
>
>
>
> betadraw
>
> An ni by natt by floor(use/keep) array of MCMC random draws of unit-level
> multinomial logit model parameter estimates.
>
> betadraw.c
>
> An ni by natt by floor(use/keep) array of constrained MCMC random draws of
> unit-level multinomial logit model parameter estimates.
>
>
>
> If I do:
>
> apply(out$betadraw.c,c(1,2),mean)
>
> then what is the difference between the result and the file written to
Disk?
>
> One difference I noticed:
>
> apply(out$betadraw.c, c(1,2),mean) contains as many columns as parameters,
> not as many columns as the number of levels.
>
> What else?
>
>
>
> Thank you very much!
>
>