1.5 Outro

1.5.1 Remarks

In supervised learning, with each input point, there’s an associated reference output value.

Learning a model = constructing a function that approximates (minimising some error measure) the given data.

Regression = the output variable $$Y$$ is continuous.

We studied linear models with a single independent variable based on the least squares (SSR) fit.

In the next part we will extend this setting to the case of many variables, i.e., $$p>1$$, called multiple regression.

1.5.2 Further Reading

Recommended further reading:

• (James et al. 2017: Chapters 1, 2 and 3)

Other:

• (Hastie et al. 2017: Chapter 1, Sections 3.2 and 3.3)