<aside> 💡
Regularisation techniques help improve the performance of logistic regression models by preventing overfitting, where the model learns the noise in the data rather than the true relationship.
</aside>
L1 regularisation, or Lasso, adds a penalty proportional to the absolute value of the coefficients, encouraging some coefficients to become zero, which simplifies the model.
In simpler terms, L1 regularisation adds or removes features from the model by making some coefficients exactly zero.
L2 regularisation, or Ridge, adds a penalty proportional to the square of the coefficients, shrinking them towards zero but not eliminating them.
Both techniques strike a balance between model fit and complexity, improving generalisability of our models to new data.
Next element: