Creating correct and capable classifiers
summary
A practical workflow that starts with a concrete baseline and ends with explainable models the talk claims reduces guesswork and reveals real bugs: use pandas_profiling reveals 20% missing age imputed to 28 spike, build a dummy classifier ~60% baseline on Titanic, move to a random forest 76–90% via repeated k‑fold whose ROC AUC beats logistic, and use Shapley explains individual predictions plus YellowBrick visualizations to expose which features drive errors and whether fixes actually improve outcomes.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 4:16Event etiquette
- 7:00Pandas profiling for data exploration
- 11:33Baseline dummy classifier
- 19:38Diagnosing classifier errors
- 25:15Shapley explanations for model insights
- 32:23Debugging machine learning
- 34:10Q&A
related talks
42mCreating correct and capable classifiers
27mUsing Machine Learning to solve a classification problem with scikit-learn
01h 33mUnderstanding and diagnosing your machine-learning models
44mHandling predictive uncertainty in Machine Learning
57mHow To Evaluate and Improve Machine Learning Models
39mBeyond Bias: Crafting Fairness in Pythonic Machine Learning Systems
42mMachine Learning Crash Course
39mKeynote 3: How did you know? Explaining Black Box Model Predictions in Python