The learning algorithm is how we search for the best weight vector (w). This search is an optimization problem looking for the hypothesis that optimizes an error/cost measure.

So, the cost or the loss function of the model is going to tell us our model is bad, and we need to minimize this function. There are different loss or cost criteria that you can follow. In this implementation, we are going to use mean squared error (MSE) as a loss function.

To accomplish the task of minimizing the loss function, we are going to use the gradient descent algorithm.