Create a modeling pipeline

Create a modeling pipeline

Usage

tl_pipeline(
  data,
  formula,
  preprocessing = NULL,
  models = NULL,
  evaluation = NULL,
  ...
)

Arguments

data

A data frame containing the data

formula

A formula specifying the model

preprocessing

A named list of preprocessing switches, each TRUE or FALSE: impute_missing (default TRUE) replaces missing predictor values with the training median or mode; standardize (default TRUE) centres and scales numeric predictors where that leaves the model the formula describes unchanged. It leaves alone any column the formula uses inside a function call such as log(), poly() or offset(), every column when the formula has no intercept, and the columns of an interaction whose lower-order terms are not all in the formula; dummy_encode (default TRUE) only records that categorical predictors are encoded by each model's fitting function, and cannot be set to FALSE.

models

A list of models to train

evaluation

A list of evaluation criteria

...

Not used. Anything passed here is an error, so a misspelt argument such as evalution is reported rather than ignored.

Value

A tidylearn_pipeline object (S3 list) with components $formula, $data, $preprocessing, $models, $evaluation, and $results (initially NULL; populated after tl_run_pipeline).

Examples

# \donttest{
pipe <- tl_pipeline(iris, Species ~ .,
  models = list(tree = list(method = "tree")))
print(pipe)
#> Tidylearn Pipeline
#> =================
#> Formula: Species ~ . 
#> Data: 150 observations, 5 variables
#> Preprocessing: impute_missing, standardize, dummy_encode 
#> Models: tree 
#> Evaluation:  cv (5 folds)
#> Metrics: accuracy, precision, recall, f1, auc 
#> Best metric: f1 
# }