September 18, 2024

Part 5.1: Introduction to Regularization: Ridge and Lasso
t81_558_class_05_1_reg_ridge_lasso.py

The dataset auto-mpg.csv is used, where
- mpg is the dependent variable
- 'cylinders', 'displacement', 'horsepower', 'weight', 'acceleration', 'year', 'origin' are the predictor variables.

Note that this is an application of
"Multidimensional Linear Regression" or "Multiple Linear Regression"
where multiple predictor variables are used.

y = β₀ + β₁x₁ + β_nx_n + ε

We are looking for the coefficients β_i that lead to the minimal model error.

https://www.scribbr.com/statistics/multiple-linear-regression/

September 19, 2024

Part 5.2: Using K-Fold Cross-validation with Keras

scikit-learn offers KFold and StratifiedKFold.
- use KFold for Regression problems
- use StratifiedKFold for Regression problems
because in classification problems, using KFold could lead to a drift in balance of classes,
StratifiedKFold prevents unintended shift.

Date