Machine Learning for Developers

Determination of the cost function

As with all machine learning techniques, the process of learning depends on a minimized loss function, which shows us how right or wrong we are when predicting an outcome, depending on the stage of learning we are in.

The most commonly used cost function for linear regression is called least squares.

Let's define this cost function taking, for simplification a 2D regression, where we have a list of number tuples (x₀, y₀), (x₁, y₁) ... (x_n, y_n) and the values to find, which are β₀  and β₁. The least squares cost function in this case can be defined as follows:  

Least squares function for a linear equation, ...