k-NN prediction
Majority vote among the k nearest neighbours.
← Machine Learning learning path0%
12
Supervised learning
Nearest-Neighbor & Probabilistic Methods
Predicting from similar examples, and classifying with Bayes' theorem.
Not started yet
Start Quiz mastery0%
What you’ll take away
- k-NN predicts from the k closest training points; k controls smoothness.
- Scale features before any distance-based method.
- Naïve Bayes assumes conditional independence and still classifies surprisingly well.
- Probabilistic classifiers pick the class with the highest posterior.
Major concepts (syllabus)
k-NN, distance measures, Naïve Bayes, probabilistic classification
Leads to 13. Unsupervised Learning
Lessons
- k-Nearest NeighborsVote (or average) among the k closest training points.9 min
- Distance measures for k-NNManhattan, Euclidean and others, and why features must be scaled.8 min
- Naïve BayesBayes' theorem with conditionally independent features.9 min
- Probabilistic classificationChoosing the class with the highest posterior probability.8 min
Key formulas
Naïve Bayes
Prior × product of per-feature likelihoods.
Key terms
- k-NN
- Predicting from the labels of the k nearest training points.
- Naïve Bayes
- Classifier applying Bayes' theorem with conditionally independent features.
Practice this topic5 questions · earn XP and build your streak