Problems with using least squares for classification
							logistic regression
						least squares regression
				If the right answer is 1 and the model says 1.5, it loses, so it changes the boundary to avoid being “too correct”