Visual Diagnostics for More Effective Machine Learning
summary
Demonstrates a concrete library and patterns: the Yellowbrick Visualizer that plugs into the scikit‑learn API, producing feature, classification, clustering, and regression diagnostics. Shows recursive feature elimination plots, cross‑validation variability heat maps, manifold embeddings (PCA, t‑SNE, UMAP) and learning/validation curves, and reports where regularization or normalization actually moves metrics. Gives examples of what changed the model, how many features proved unnecessary, and which visual checks find heteroscedasticity or slow hyperparameter regimes.
AI-written summary · about
chapters (13)
- 0:00Visual diagnostics intro
- 1:06Custom code to estimator APIs
- 7:24Model search challenge
- 9:28Introducing Yellowbrick
- 10:10Feature analysis in high dim
- 13:46Radial and parallel coordinates
- 17:40PCA and embeddings
- 19:49Feature selection and complexity
- 22:17Cross-validation visuals
- 24:46Clustering and tuning diagnostics
- 26:51Visualizers in ML workflows
- 28:38Open-source goals
- 30:07Q&A
related talks
39mVisual diagnostics for more informed machine learning
39mYellowbrick: Steering Machine Learning with Visual Transformers
38mBuilding a Visualizer in Yellowbrick
01h 33mUnderstanding and diagnosing your machine-learning models
43mVisualising Machine Learning Models
35mDeep Learning Model Interpretability for Computer Vision based Models
22mVisualizing machine learning algorithms in the Jupyter notebook
02h 23mEffective Data Visualization