Linear Regression (page 2 of 7)

The linear regression algorithm follows a process of gradient descent (full, stochastic, or batch) to minimize a cost function associated with the losses across each observation. With certain linear regression algorithms in sklearn, we have the ability to specify hyper-parameters for the learning rate and other details related to the gradient descent process. The base LinearRegression algorithm does not give us any of these hyper-parameter options. Therefore, it may not be an efficient algorithm for complex data.