R/methods_mclogit.R

In marginaleffects: Predictions, Comparisons, Slopes, Marginal Means, and Hypothesis Tests

#' @rdname get_group_names
#' @export
get_group_names.mblogit <- function(model, type, ...) {
    out <- get_predict(model, type = type)
    if ("group" %in% colnames(out)) {
        out <- unique(out$group)
    } else {
        out <- "main_marginaleffects"
    }
    return(out)
}

#' @include sanity_model.R
#' @rdname sanitize_model_specific
#' @export
sanitize_model_specific.mblogit <- function(
    model,
    calling_function = "marginaleffects",
    ...
) {
    if (calling_function == "marginaleffects") {
        variables <- insight::find_variables(model, flatten = TRUE, verbose = FALSE)
        dat <- insight::get_data(model)
        dat <- dat[, intersect(variables, colnames(dat))]
        flag <- any(sapply(dat, is.character))
        if (isTRUE(flag)) {
            stop(
                "Cannot compute marginal effects for models of class `mblogit` when the data includes character variables. Please convert character variables to factors in the dataset before fitting the model, and call `marginaleffects` again.",
                call. = FALSE
            )
        }
    }
    return(model)
}


#' @rdname get_predict
#' @export
get_predict.mblogit <- function(
    model,
    newdata = insight::get_data(model),
    type = "response",
    ...
) {
    out <- suppressMessages(
        get_predict.multinom(
            model = model,
            newdata = newdata,
            type = type,
            ...
        )
    )
    return(out)
}


#' @rdname get_coef
#' @export
get_coef.mblogit <- function(model, ...) {
    # get_coef.default uses `insight::get_parameters`, but does not combine
    # Response and Parameter columns. This also matches naming scheme from
    # `vcov`
    stats::coef(model)
}