compute_model_weights: Compute (log) weights of component models based on an...

In reichlab/xgbstack: Flexible model stacking via gradient boosting for models that output predictive distributions.

Description

Format "complete" is only partially implemented.

Usage

compute_model_weights(xgbstack_fit, newdata, ntreelimit, log = FALSE,
  format = "bare")

Arguments

xgbstack_fit	a fit xgbstack object
newdata	new x data
ntreelimit	how many boosting iterations worth of trees to use
log	boolean: return log of weights or original weights?
format	string, either "bare" to return an nrow(newdata) by num_models matrix of (log) weights, or "complete" to return a data frame with nrow(newdata) times num_models data frame with both weights and log_weights as well as parameter values and newdata

Value

(log) weights in the format determined by format

reichlab/xgbstack documentation built on May 27, 2019, 4:54 a.m.