Elastic Net Regression with Python & Sklearn
Python
Elastic Net is a linear regression algorithm that uses a combination of L1 and L2 regularisation so can be thought of as a blend of Lasso and Ridge regression algorithms.
In the code snippet below we initialise and fit a model that combines an equal blend of L1 and L2 regularisation before evaluating the model against test data.
1| from sklearn.linear_model import ElasticNet 2| from sklearn.metrics import mean_squared_error, mean_absolute_error, max_error, explained_variance_score, mean_absolute_percentage_error 3| 4| # initialise & fit Elastic Net regression model with alpha set to 0.1 5| # and 50/50 blend between l1 and l2 regularisation 6| model = ElasticNet(alpha=0.1, l1_ratio=0.5) 7| model.fit(X_train, y_train) 8| 9| # make prediction for test data & evaluate performance 10| y_pred = model.predict(X_test) 11| print('RMSE:',mean_squared_error(y_test, y_pred, squared = False)) 12| print('MAE:',mean_absolute_error(y_test, y_pred)) 13| print('MAPE:',mean_absolute_percentage_error(y_test, y_pred)) 14| print('Max Error:',max_error(y_test, y_pred)) 15| print('Explained Variance Score:',explained_variance_score(y_test, y_pred))
133
121
117
109