X.nl2: function to estimate 2-level nested logit model
In xf-pan/dicmo: Discrete Choice Modeling

Description Usage Arguments Author(s)

View source: R/X.nl2.R

This function could estimate 2-level nested logit model along with converting the data from a wide format to a long format and code the categorical attributesallow allow. In detail, it allows to estimate interaction effects between attributes and alternative-specific parameters.

X.nl2(
  data,
  choice,
  alts,
  attrs,
  nest,
  nest_uni = TRUE,
  attr_coding = NULL,
  attr_level = NULL,
  interact = NULL,
  avi = NULL,
  method = "BFGS",
  estimator = TRUE,
  param_fixed = NULL,
  param_start = NULL
)

`data`	A tibble, input data, wide format.
`choice`	A character, name of column indicating individuals' choice.
`alts`	A vector of characters, names of all alternatives, including the none-option if any.
`attrs`	A list contains three slices, whose elements have to be non- negative integers. The first one is a tibble, named "attrs_alts", indicating the alternative-specific attributes (excluding ASCs); the second one, named "asc", is a vector indicating the ASCs, and the third one, named "context", is a tibble indicating the context variables (includig individuals' socio-demographics). The column names in all tibbles represent the name of attributes, or context variables. All have same numbers of rows (elements) with the length of argument "alts". The element in tibbles/vector indicates if the attribute, constant or context variable is alternative-specific: "0" means the attribute, constaant or context variable is not available for the alternative (based on the sequence of alternative in the argument "alts"); except "0", if some elements in one column have a same value, then the corresponding alternatives have generic parameter in terms of this attribute, constant or context variable.
`nest`	A list, indicating how many nests at all and which alternatives are in the same nest.
`nest_uni`	A logical, indicating if the inclusive values for the nests are the same or not. Default = TRUE, means the inclusive values are the same.
`attr_coding`	A vector of character, names of categorical attributes. Default = NULL, which means all attributes are continuous.
`attr_level`	A list of vector of character, representing the levels of all categorical attributes. The first element of a vector is set as base level, others are set as 1, 2, 3...in sequence. This argument is only relevent when argument "attr_coding" is not NULL. Default = NULL.
`interact`	A vector of character, name of attributes' interaction, connected by "*". Default = NULL.
`avi`	A character, name of column indicating if an alternative is available to individuals. Default = NULL, indicating all alternatives are available to all respondents. Each alternative should have such a column, for example, avi = "available" then the column's name for an alternative (e.g. the alternative's name is "car") should be "available:car" or "car:avilable". If this parameter is NULL, then those columns are not necessary. If this parameter is not NULL, then in such as column, the element should be 0 if the alternative is not available otherwise 1.
`method`	A character, passed to the function maxLik() in "maxLik" package. It indicates the method used in maximum likelihood estimation. Default = "BFGS".
`estimator`	A argument in the function maxLik() from "maxLik" package. Here this argument could only be either TRUE or "BHHH"/"bhhh". Default = TRUE.
`param_fixed`	A vector of characters, passed to the function maxLik() in "maxLik" package. It indicates which parameters are fixed. Default = NULL.
`param_start`	A vector of numbers, passed to the function maxLik() in "maxLik" package. It indicages the initial values of parameters. Default = NULL.