RunConsiderationSet: Consideration Set Model
In hollicam/DiscreteChoiceModels: Various models for marketing and social research

Description Usage Arguments Value Examples

The consideration set model takes data with the format of each row representing a choice for some person. First, the model constructs a set of choices in consideration for each person via a binary logit model (the Selection step), then selects choices from consideration set via a multinomial logit model (the Consideration step).

1
2
3

RunConsiderationSet(YObservedInput, XConsidInput, XSelectInput, NumOpts,
  ASCConsid = 0, ASCSelect = 0, IncludedInput = 0 * YObservedInput + 1,
  meth = "L-BFGS-B", Tolerance = 1e-05)

`YObservedInput`	A vector of observed choices acting as the response variable for all subjects, each row representing a unique choice and the value representing how many times it was chosen.
`XConsidInput`	A vector or matrix of factors that influences what choices are to be chosen from the consideration set, i.e. for second step.
`XSelectInput`	A vector or matrix of factors that influences what choices are to be included in the consideration set, i.e. for first step.
`NumOpts`	Number of options for consideration set. Due to the nature of the model, NumOpts needs to be a factor of the total number of rows, such that every n (value of NumOpts) rows represent a set of observed choices for one subject.
`ASCConsid`	Binary flag to indicate if users want ASC, i.e. intercept in the Consideration step; 0, i.e. "No", by default.
`ASCSelect`	Binary flag to indicate if users want ASC, i.e. intercept in the Selection step; 0, i.e. "No", by default.
`IncludedInput`	A vector or matrix of which choices are available for constructing the model; every choice is available, i.e. a vector of 1's for all YObservedInput, by default. (A value of 0 excludes a particular choice.)
`meth`	Algorithm used as method for optimization. Methods available are "L-BFGS-B", "BFGS", "CG", "SANN", "Brent"; "L-BFGS-B" by default.
`Tolerance`	(Valid only when `meth="L-BFGS-B"`) Controls the convergence of the "L-BFGS-B" method; 1e-5 by default. Convergence occurs when reduction in the objective is smaller in magnitude than this specified factor of machine tolerance.

A list object containing the following:

Likelihood for the consideration set model.
Likelihood for the logit model.
Coefficients for all covariates in Consideration and Selection input. If ASCConsid / ASCSelect is set equal to 1, the output will include ASC in Consideration / Selection.

data(CFS5-1)
attach(CFS5-1)
RunConsiderationSet(V1, cbind(V2,V3,V4), cbind(V2,V3,V4), 5, 1, 1,Tolerance=1e-6)

data(CFS6-1)
attach(CFS6-1)
RunConsiderationSet(V1, cbind(V2,V3,V4), cbind(V2,V3,V4), 6, 1, 1)

data(CFS5-2)
attach(CFS5-2)
RunConsiderationSet(V1, cbind(V2,V3,V4), cbind(V2,V3,V4), 5, 1, 1, IncludedInput=V5, meth="BFGS")

data(CFS6-2)
attach(CFS6-2)
RunConsiderationSet(V1, cbind(V2,V3,V4), cbind(V2,V3,V4), 6, 1, 1, IncludedInput=V5, meth="BFGS")