Algorithm¶
Logistic Regression¶
Support Vector Machine, SVM¶
K-Nearest Neighbors,KNN¶
supervised learning
For classification:
Given a test instance i, find the k closest neighbors and their labels
Predict i’s label as the majority of the labels of the k nearest neighbors
For regression:
Given a test instance i, find the k closest neighbors and their value
Predict i’s value as the mean of the value of the k nearest neighbors
Could be used directly without training. “Training” process is building a index for searching neighbors faster.
small k-> overfitting
large k-> underfitting
K-Means¶
clustering