It's Not Magic: Explaining Classification Algorithms
summary
Survey of about ten concrete classifiers with code and concrete tradeoffs: linear discriminant and logistic regression give linear boundaries and coefficient-based importance, Scikit Learn snippets show fit and predict, SVMs demonstrate the RBF kernel and margin maximization via a kernel trick (squaring feature example), K‑nearest neighbors shows how K smooths jagged boundaries, decision trees produce literal flowcharts, ensembles (bagging, Random Forest, Extra Trees, boosting) combine models to reduce overfit while trading compute and update/parallelization costs.
AI-written summary · about
chapters (11)
- 0:00Classification algorithms intro
- 5:59Linear discriminant analysis
- 12:03Logistic regression basics
- 13:08Support vector machines & kernels
- 18:44K nearest neighbors
- 22:18Decision trees & modeling
- 25:27Ensemble methods: bagging & forests
- 30:19Boosting & voting classifiers
- 33:12Choosing the right algorithm
- 40:13Model quirks & practical tips
- 42:01Summary & Q&A
related talks
45mFrom Ancient Greeks to Modern Geeks - Basic Machine Learning Algorithms
40mCART: Not only Classification and Regression Trees
30mMachine Learning Techniques for Class Imbalances & Adversaries
42mDistribution-Specific Analysis of Nearest Neighbor Search and Classification
01h 33mUnderstanding and diagnosing your machine-learning models
28mClassification Through Regression:Unlock the True Potential of Your Labels
37mCreating correct and capable classifiers
40mTools and Tricks from a Pragmatic Data Scientist