Lasso Regression
Just like Ridge Regression it uses a penalty / Regularisation term to regularise the slope of the Regression line and performs Feature Selection.
where is the slope.
Minimize the sum of least squares and a penalty or Regularisation term.
Used to prevent Overfitting.
Also can exclude useless variables/features completely by minimizing their weights (slope) right down to zero. This allows us to do Feature Selection and also infer some kind of Feature Importance from the weight plots:
In this case the windspeed and humidity dont seem to play such a huge role. To find the optimal value for lambda use K-Fold Cross Validation.
Classification
This method also works for Logistic Regression Classification.
Explainability
Just like in Linear Regression with the added benefit of automatic Feature Selection and Regularisation.