Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!
Get Started for Free

createX

Create X Matrix for Use in Multinomial Logit and Probit Routines


Description

createX makes up an X matrix in the form expected by Multinomial Logit (rmnlIndepMetrop and rhierMnlRwMixture) and Probit (rmnpGibbs and rmvpGibbs) routines. Requires an array of alternative-specific variables and/or an array of "demographics" (or variables constant across alternatives) which may vary across choice occasions.

Usage

createX(p, na, nd, Xa, Xd, INT = TRUE, DIFF = FALSE, base=p)

Arguments

p

integer number of choice alternatives

na

integer number of alternative-specific vars in Xa

nd

integer number of non-alternative specific vars

Xa

n x p*na matrix of alternative-specific vars

Xd

n x nd matrix of non-alternative specific vars

INT

logical flag for inclusion of intercepts

DIFF

logical flag for differencing wrt to base alternative

base

integer index of base choice alternative

Note: na, nd, Xa, Xd can be NULL to indicate lack of Xa or Xd variables.

Value

X matrix of dimension n*(p-DIFF) x [(INT+nd)*(p-1) + na].

Note

rmnpGibbs assumes that the base alternative is the default.

Author(s)

Peter Rossi, Anderson School, UCLA, perossichi@gmail.com.

References

For further discussion, see Bayesian Statistics and Marketing by Rossi, Allenby, and McCulloch.
http://www.perossi.org/home/bsm-1

See Also

Examples

na=2; nd=1; p=3
vec = c(1, 1.5, 0.5, 2, 3, 1, 3, 4.5, 1.5)
Xa = matrix(vec, byrow=TRUE, ncol=3)
Xa = cbind(Xa,-Xa)
Xd = matrix(c(-1,-2,-3), ncol=1)
createX(p=p, na=na, nd=nd, Xa=Xa, Xd=Xd)
createX(p=p, na=na, nd=nd, Xa=Xa, Xd=Xd, base=1)
createX(p=p, na=na, nd=nd, Xa=Xa, Xd=Xd, DIFF=TRUE)
createX(p=p, na=na, nd=nd, Xa=Xa, Xd=Xd, DIFF=TRUE, base=2)
createX(p=p, na=na, nd=NULL, Xa=Xa, Xd=NULL)
createX(p=p, na=NULL, nd=nd, Xa=NULL, Xd=Xd)

bayesm

Bayesian Inference for Marketing/Micro-Econometrics

v3.1-4
GPL (>= 2)
Authors
Peter Rossi <perossichi@gmail.com>
Initial release
2019-10-14

We don't support your browser anymore

Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.