ML module Week 2 Topic 3: Linear regression

[EricThomson]

Once the students appreciate the world of ML, and the general scikit-learn ecosystem, let's dig into detail into linear regression, the goal is to train and evaluate a regression model using scikit-learn.

1. Discussion of linear regression: what is it, how is it connected to linear correlation from week 1.
2. Use scikit learn to create model, do model fit, and model prediction (using the standard three-part scikit-learn api). Lots of data points with noisy data.
3. Evaluate model fit with visualizations, and metrics like MSE and R2 (explain these metrics)
4. Break data into training/test data to illustrate generalization
   Use the extremelyl important test_train_split function to break data into test/train data and explain why we are doing this (to avoid overfitting!)
   Show polynomial regression to illustrate overfitting (here let's do it with few data points to illustrate the point of lack of generalization).