By applying a threshold to probability

By minimizing Euclidean distance

By optimizing the margin

By counting nearby neighbors

To avoid overfitting the model

To improve data collection speed

To increase the number of features

To reduce the size of the dataset

It assumes feature independence

It stores data and predicts at query time

It stores data and trains slowly using iterations

It ignores feature scaling

The probability of correct classification

The distance to the closest points of each class

The average distance between all points

The number of training samples used in each vector

Logistic Regression

Linear Regression

Mean Squared Error

K-Means Clustering