parsnip A tidy model interface
summary
A new R package unifies model specification from estimation by letting users declare a model type (e.g. linear or logistic) and swap engines like glmnet, Stan, Spark or keras without rewriting code. It returns standardized tibble .pred outputs that always match input rows, supports glmnet’s full lambda path via multi_predict list-column, and computes engine-aware parameters with data descriptor expressions (e.g. mtry as floor(0.75 * num_predictors)). The project builds on ~250 modeled interfaces and shows few GitHub issues so far.
AI-written summary · about
chapters (8)
- 0:00Intro
- 0:58Modeling in R - User-facing Problems
- 4:14parsnip
- 9:31Example: Linear Regression
- 11:47Prediction
- 16:44Data Descriptors
- 18:03Specifying mtry
- 19:30Next Steps
related talks
15mThe tidysynthesis R package
57mGood Practices for Applied Machine Learning
18mtidymodels: Adventures in Rewriting a Modeling Pipeline
21mTidypredict with recipes, turn workflow to SQL, spark, duckdb and beyond
18mCensored - Survival Analysis in Tidymodels
50mLifting the Curtain on Machine Learning
15mInterpretable Machine Learning with R
32mAn Introduction to 'metalearners', a Flexible MetaLearner Library in Python