Tidymodels

Tidymodels and tune

• tidymodels, tune
• https://www.tidymodels.org/
• Tidy Modeling with R
• *Translations of the ISLR labs into using the tidymodels set of packages
• Caret for Survival Analysis?
• Survival Methods in tidymodels, 2021/11/2
• censored package. censored is a “parsnip-adjacent” packages with model definitions for censored regression and survival analysis models.
• A Gentle Introduction to tidymodels
• Julia Silge Handle class imbalance in climbing expedition data with tidymodels
• Parallel processing with tune
• Tutorial on tidymodels for Machine Learning
• Julia Silge Lasso regression using tidymodels and #tidytuesday data for the office
  • Explore the data: transmute(), mutate(), select(), count(), add_count(), filter(), select(), pivot_wider(), distinct(), pivot_longer(), separate_rows(), add_count(), filter(), distinct(), mutate(), pivot_wider(), inner_join(), janitor::clean_names(), geom_boxplot()
  • Train a model: initial_split(), training(), testing(), recipe(), update_role(), step_zv(), step_normalize(), prep(), linear_reg(), set_engine(), workflow(), add_recipe(), add_model(), fit(), pull_workflow_fit(), tidy()
  • Tune lasso parameters: bootstraps(), linear_reg(), set_engine(), grid_regular(), registerDoParallel(), tune_grid(), add_model(), collect_metrics(), geom_errorbar(), geom_line(), facet_wrap(), scale_x_log10(), theme(), select_best(), finalize_workflow(), add_model(), pull_workflow_fit(), geom_col(), scale_x_continuous(), last_fit(), collect_metrics()
• Tidymodel and glmnet - Jun
• Tidymodels: tidy machine learning in R - Rebecca Barter
• Tidy Model Stacking 2021 John M. Chambers Statistical Software Award.
• Julia Silge Explore art media over time in the #TidyTuesday Tate collection dataset
• Julia Silge Supervised Machine Learning for Text Analysis in R is now complete,
  • Online book https://smltar.com/
• Intro to tidymodels with NFL data
• Julia Silge Multiclass predictive modeling for #TidyTuesday NBER papers. penalty was tuned over a grid of possible penalty parameters by using tune_grid(), doParallel and select_by_one_std_err(). There is also a confusion matrix plot and ROC curves for each class.
  • Build and tune a model: initial_split(), training(), test(), vfold_cv(), recipe(), multinom_reg(), set_mode(), set_engine(), workflow(), grid_regular(), tune_grid(), autoplot(), show_best()
  • Choose and evaluate a final model: select_by_one_std_err(), finalize_workflow(), last_fit(), collect_metrics(), conf_mat(), collect_predictions(), roc_curve(), extract_workflow(), predict()
• TidyModels or CARET – how they compare?
  • Explore data: separate(), mutate(), filter(), select(), geom_point(), geom_smooth, scale_color_viridis_c(), geom_boxplot(), geom_jitter(), theme()
  • Build and fit a workflow set: filter(), initial_split(), training(), testing(), vfold_cv(), recipe(), step_other(), step_dummy(), step_bs(), rand_forest(), set_mode(), set_engine(), linear_reg(), workflow_set(), registerDoParallel(), workflow_map()
  • Evaluate workflow set: autoplot(), collect_metrics(), extract_workflow(), fit(), tidy(), arrange()
• LASSO models in tidymodels from STAT 253: Statistical Machine Learning
  • Lasso model spec: linear_reg(), set_args(), set_engine(), set_mode()
  • Recipe with standardization: recipe(), step_nzv(), step_novel(), step_normalize(), step_dummy()
  • Workflow: workflow(), add_recipe(), add_model()
  • Fit model:: fit()
  • Examine the model for each lambda: plot( extract_fit_parsnip()) , pluck()
  • Identifying the best Lasso model: vfold_cv(), grid_regular(), tune_grid()
  • Visualize model evaluation metrics from tuning: autoplot(), theme_classic()
  • Summarize model evaluation metrics: collect_metrics(), filter(), select()
  • Fit final model: finalize_workflow(), fit(), tidy()
• TidyModels hands on for the NHS-R community – building a supervised ML classification model from scratch
• Closing out our year with a Q4 2021 tidymodels update

stacks

https://stacks.tidymodels.org/, JSM 2021 John Chambers Award Recipient