40 Algorithms Every Programmer Should Know

Evaluating the regressors

The linear line that we drew is an approximation of the relationship between the dependent and independent variables. Even the best line will have some deviation from the actual values, as shown:

A typical way of quantifying the performance of linear regression models is by using Root Mean Square Error (RMSE). This calculates the standard deviation of the errors made by the trained model mathematically. For a certain example in the training dataset, the loss function is calculated as follows:

Loss (ý⁽ⁱ⁾, y⁽ⁱ⁾) = 1/2(ý^(i)- y⁽ⁱ⁾)²

This leads to the following cost function, which minimizes the loss of all of the examples ...