_targets.R

In gouthaman87/ameshouseR: What the Package Does (One Line, Title Case)

library(targets)

file_functions = list.files(path = "R", full.names = TRUE)

sapply(file_functions, source)

tar_option_set(
  packages = c("tidyverse", "tidymodels", "agua")
)

list(
  # Read Data ----
  tar_target(name = raw_data,
             read_from_path(path = "data/ames.rda"),
             priority = 1),

  tar_target(name = param_file,
             "params.yaml",
             format = "file",
             priority = 0.9),

  tar_target(name = params,
             yaml::read_yaml(param_file),
             priority = 0.8),

  # Split Data ----
  tar_target(name = train_test_split,
             data_split(DF = raw_data,
                        prop = 0.8,
                        resp = "Sale_Price"),
             priority = 0.7),

  # Recipe ----
  tar_target(name = model_recipe,
             data_recipe(DF = rsample::training(train_test_split),
                         features = params[["features"]]),
             priority = 0.6),

  # Model workflow sets ----
  tar_target(name = models_workflowset,
             model_workflow(DF = rsample::training(train_test_split),
                            model_name = c("drf"),
                            features = params[["features"]])),

  # Train the models ----
  tar_target(name = fitted_models,
             model_fit(splt = train_test_split,
                       type_of_resample = "cv",
                       model_set = models_workflowset)),

  # # Asses the models ----
  # tar_target(name = train_metrics,
  #            tune::collect_metrics(fitted_models, summarize = FALSE) |>
  #              dplyr::filter(.metric == "rmse")),
  #
  # Predict Values ----
  tar_target(name = predict_data,
             predict_values(DF = rsample::testing(train_test_split),
                            model_fit = fitted_models))
)