← PreviousYou are hereReadLinear regression with many featuresPredict with a weighted sum of features, and find the weights by least squares in one step or by gradient descent on the same bowl, with the same answer.