plotLR¶
Purpose¶
Generates summary plot of parameter path and mse path over lambda values.
Format¶
-
plotLR(mdl[, mse_test])¶ Parameters: - mdl (struct) – A filled instance of either a
lassoModelorridgeModelstructure. - mse_test (Nx1 vector) – Optional, vector of MSE values for test data.
- mdl (struct) – A filled instance of either a
Examples¶
new;
library gml;
// Specify dataset with full path
dataset = getGAUSSHome("pkgs/gml/examples/qsar_fish_toxicity.csv");
// Split data into training sets without shuffling
shuffle = "False";
{ y_train, y_test, X_train, X_test } = trainTestSplit(dataset, "LC50 ~ . ", 0.7, shuffle);
// Declare 'mdl' to be an instance of a
// lassoModel structure to hold the estimation results
struct lassoModel mdl;
// Estimate the model with default settings
mdl = lassoFit(y_train, X_train);
/*
** Prediction for test data
*/
y_hat = lmPredict(mdl, X_test);
test_mse = meanSquaredError(y_test, y_hat);
/*
** Plot results
*/
plotLR(mdl, test_mse);
This results in the following plot:
See also
Functions lmPredict(), lassoFit(), ridgeFit()
