Linear Model Coefficients

In Chapter 2, Regression, and Chapter 3, Binary Classification, you saw that linear regression models learn function parameters in the form of the following:

Figure 9.1: Function parameters for linear regression models

The objective is to find the best parameters (w1, w2 …, wn) that will get the predictions, ŷ̂, very close to the actual target values, y. So, once you have trained your model and are getting good predictive performance without much overfitting, you can use these parameters (or coefficients) to understand which variables largely impacted the predictions. If a coefficient is close to 0, this means the related feature didn't impact much the outcome. On the other hand, if it is quite high (positively or negatively), it means its feature is influencing the prediction outcome vastly.

Let's take the example of the following function: 100 + 0.2 * x1 + 200 * x2 - 180 * x3. The...