stack_ml_models: Creates super-learner by stacking differnt ML models

In vidarsumo/sumots: Helper functions for Sumo Analytics

Description

Creates super-learner by stacking differnt ML models

Usage

stack_ml_models(
  submodels_resamples,
  kfold,
  grid,
  models,
  parallel = TRUE,
  verbose = TRUE
)

Arguments

submodels_resamples	A modeltime table (mdl_time_tbl) object with a column containing resample results. Created with modeltime_fit_resamples() from the modeltime.resample package
kfold	Number of folds in K-fold cross validation
grid	Size of grid to tune the meta-learner. Gets passed to tune::tune_grid()
models	Which models to try for the meta-learner. Possible choices: glmnet, rf, nnet, xgboost, cubist and svm
parallel	Should the calculations be run in parallel. Defaults to TRUE
verbose	A logical for logging results as they are generated. Defaults to TRUE

Value

A list of length two with ensemble accuracy and a modeltime_table to be used for forecasting

vidarsumo/sumots documentation built on June 29, 2021, 4:23 a.m.