For greater accuracy on low- through medium-dimensional data sets, implement least-squares regression with regularization using lasso or ridge.
For reduced computation time on high-dimensional data sets, fit a regularized linear regression model using fitrlinear.
lasso | Lasso or elastic net regularization for linear models |
ridge | Ridge regression |
lassoPlot | Trace plot of lasso fit |
fitrlinear | Fit linear regression model to high-dimensional data |
predict | Predict response of linear regression model |
RegressionLinear | Linear regression model for high-dimensional data |
RegressionPartitionedLinear | Cross-validated linear regression model for high-dimensional data |
See how lasso identifies and discards unnecessary
predictors.
Lasso and Elastic Net with Cross Validation
Predict the mileage (MPG) of a car based on its weight, displacement,
horsepower, and acceleration using lasso and elastic
net.
Wide Data via Lasso and Parallel Computing
Identify important predictors using lasso and
cross-validation.
The lasso algorithm is a regularization
technique and shrinkage estimator. The related elastic net algorithm
is more suitable when predictors are highly correlated.
Ridge regression addresses the problem of multicollinearity (correlated model terms) in linear regression problems.