tidyclust - expanding tidymodels to clustering
summary
A new package tidyclust provides a tidy, unified interface for clustering inside tidymodels, implementing k‑means as a model spec with parameters like num_clusters, partition vs future extractive clustering, and engines beyond stats. Workflows and recipes run preprocessing (normalization, PCA) before fitting, centroids are extractable via extract_centroids, predictions use nearest‑centroid logic, and cluster metrics include total within‑sum‑of‑squares. Tuning uses tune_cluster over a grid; package is built and working but still early development.
AI-written summary · about
chapters (9)
- 0:00Introduction to tidymodels
- 2:25Introducing tidycluster
- 4:17Challenges in clustering
- 7:14Example data overview
- 9:00Clustering workflow
- 11:59Extracting model results
- 15:10Clustering metrics
- 16:12Tuning clustering models
- 18:17Conclusion and next steps
related talks
19mtidyclust - expanding tidymodels to clustering
35mClustering: A Guide for the Perplexed
01h 20mClustering Data
57mGood Practices for Applied Machine Learning
18mtidymodels: Adventures in Rewriting a Modeling Pipeline
47mAssessing the quality of a clustering
21mTidypredict with recipes, turn workflow to SQL, spark, duckdb and beyond
24mNetworkX - your Unexpected Assistant for Clustering Analysis