139 lines
6.3 KiB
R
139 lines
6.3 KiB
R
#' Importance of features in a model.
|
|
#'
|
|
#' Creates a \code{data.table} of feature importances in a model.
|
|
#'
|
|
#' @param feature_names character vector of feature names. If the model already
|
|
#' contains feature names, those would be used when \code{feature_names=NULL} (default value).
|
|
#' Non-null \code{feature_names} could be provided to override those in the model.
|
|
#' @param model object of class \code{xgb.Booster}.
|
|
#' @param trees (only for the gbtree booster) an integer vector of tree indices that should be included
|
|
#' into the importance calculation. If set to \code{NULL}, all trees of the model are parsed.
|
|
#' It could be useful, e.g., in multiclass classification to get feature importances
|
|
#' for each class separately. IMPORTANT: the tree index in xgboost models
|
|
#' is zero-based (e.g., use \code{trees = 0:4} for first 5 trees).
|
|
#' @param data deprecated.
|
|
#' @param label deprecated.
|
|
#' @param target deprecated.
|
|
#'
|
|
#' @details
|
|
#'
|
|
#' This function works for both linear and tree models.
|
|
#'
|
|
#' For linear models, the importance is the absolute magnitude of linear coefficients.
|
|
#' For that reason, in order to obtain a meaningful ranking by importance for a linear model,
|
|
#' the features need to be on the same scale (which you also would want to do when using either
|
|
#' L1 or L2 regularization).
|
|
#'
|
|
#' @return
|
|
#'
|
|
#' For a tree model, a \code{data.table} with the following columns:
|
|
#' \itemize{
|
|
#' \item \code{Features} names of the features used in the model;
|
|
#' \item \code{Gain} represents fractional contribution of each feature to the model based on
|
|
#' the total gain of this feature's splits. Higher percentage means a more important
|
|
#' predictive feature.
|
|
#' \item \code{Cover} metric of the number of observation related to this feature;
|
|
#' \item \code{Frequency} percentage representing the relative number of times
|
|
#' a feature have been used in trees.
|
|
#' }
|
|
#'
|
|
#' A linear model's importance \code{data.table} has the following columns:
|
|
#' \itemize{
|
|
#' \item \code{Features} names of the features used in the model;
|
|
#' \item \code{Weight} the linear coefficient of this feature;
|
|
#' \item \code{Class} (only for multiclass models) class label.
|
|
#' }
|
|
#'
|
|
#' If \code{feature_names} is not provided and \code{model} doesn't have \code{feature_names},
|
|
#' index of the features will be used instead. Because the index is extracted from the model dump
|
|
#' (based on C++ code), it starts at 0 (as in C/C++ or Python) instead of 1 (usual in R).
|
|
#'
|
|
#' @examples
|
|
#'
|
|
#' # binomial classification using gbtree:
|
|
#' data(agaricus.train, package='xgboost')
|
|
#' bst <- xgboost(data = agaricus.train$data, label = agaricus.train$label, max_depth = 2,
|
|
#' eta = 1, nthread = 2, nrounds = 2, objective = "binary:logistic")
|
|
#' xgb.importance(model = bst)
|
|
#'
|
|
#' # binomial classification using gblinear:
|
|
#' bst <- xgboost(data = agaricus.train$data, label = agaricus.train$label, booster = "gblinear",
|
|
#' eta = 0.3, nthread = 1, nrounds = 20, objective = "binary:logistic")
|
|
#' xgb.importance(model = bst)
|
|
#'
|
|
#' # multiclass classification using gbtree:
|
|
#' nclass <- 3
|
|
#' nrounds <- 10
|
|
#' mbst <- xgboost(data = as.matrix(iris[, -5]), label = as.numeric(iris$Species) - 1,
|
|
#' max_depth = 3, eta = 0.2, nthread = 2, nrounds = nrounds,
|
|
#' objective = "multi:softprob", num_class = nclass)
|
|
#' # all classes clumped together:
|
|
#' xgb.importance(model = mbst)
|
|
#' # inspect importances separately for each class:
|
|
#' xgb.importance(model = mbst, trees = seq(from=0, by=nclass, length.out=nrounds))
|
|
#' xgb.importance(model = mbst, trees = seq(from=1, by=nclass, length.out=nrounds))
|
|
#' xgb.importance(model = mbst, trees = seq(from=2, by=nclass, length.out=nrounds))
|
|
#'
|
|
#' # multiclass classification using gblinear:
|
|
#' mbst <- xgboost(data = scale(as.matrix(iris[, -5])), label = as.numeric(iris$Species) - 1,
|
|
#' booster = "gblinear", eta = 0.2, nthread = 1, nrounds = 15,
|
|
#' objective = "multi:softprob", num_class = nclass)
|
|
#' xgb.importance(model = mbst)
|
|
#'
|
|
#' @export
|
|
xgb.importance <- function(feature_names = NULL, model = NULL, trees = NULL,
|
|
data = NULL, label = NULL, target = NULL){
|
|
|
|
if (!(is.null(data) && is.null(label) && is.null(target)))
|
|
warning("xgb.importance: parameters 'data', 'label' and 'target' are deprecated")
|
|
|
|
if (!inherits(model, "xgb.Booster"))
|
|
stop("model: must be an object of class xgb.Booster")
|
|
|
|
if (is.null(feature_names) && !is.null(model$feature_names))
|
|
feature_names <- model$feature_names
|
|
|
|
if (!(is.null(feature_names) || is.character(feature_names)))
|
|
stop("feature_names: Has to be a character vector")
|
|
|
|
model_text_dump <- xgb.dump(model = model, with_stats = TRUE)
|
|
|
|
# linear model
|
|
if (model_text_dump[2] == "bias:"){
|
|
weights <- which(model_text_dump == "weight:") %>%
|
|
{model_text_dump[(. + 1):length(model_text_dump)]} %>%
|
|
as.numeric
|
|
|
|
num_class <- NVL(model$params$num_class, 1)
|
|
if (is.null(feature_names))
|
|
feature_names <- seq(to = length(weights) / num_class) - 1
|
|
if (length(feature_names) * num_class != length(weights))
|
|
stop("feature_names length does not match the number of features used in the model")
|
|
|
|
result <- if (num_class == 1) {
|
|
data.table(Feature = feature_names, Weight = weights)[order(-abs(Weight))]
|
|
} else {
|
|
data.table(Feature = rep(feature_names, each = num_class),
|
|
Weight = weights,
|
|
Class = seq_len(num_class) - 1)[order(Class, -abs(Weight))]
|
|
}
|
|
} else { # tree model
|
|
result <- xgb.model.dt.tree(feature_names = feature_names,
|
|
text = model_text_dump,
|
|
trees = trees)[
|
|
Feature != "Leaf", .(Gain = sum(Quality),
|
|
Cover = sum(Cover),
|
|
Frequency = .N), by = Feature][
|
|
, `:=`(Gain = Gain / sum(Gain),
|
|
Cover = Cover / sum(Cover),
|
|
Frequency = Frequency / sum(Frequency))][
|
|
order(Gain, decreasing = TRUE)]
|
|
}
|
|
result
|
|
}
|
|
|
|
# Avoid error messages during CRAN check.
|
|
# The reason is that these variables are never declared
|
|
# They are mainly column names inferred by Data.table...
|
|
globalVariables(c(".", ".N", "Gain", "Cover", "Frequency", "Feature", "Class"))
|