To determine a good lasso-penalty strength for a linear classification model that uses a logistic regression learner, compare distributions of *k*-fold margins.

Load the NLP data set. Preprocess the data as in Estimate k-Fold Cross-Validation Margins.

Create a set of 11 logarithmically-spaced regularization strengths from $$1{0}^{-8}$$ through $$1{0}^{1}$$.

Cross-validate a binary, linear classification model using 5-fold cross-validation and that uses each of the regularization strengths. Optimize the objective function using SpaRSA. Lower the tolerance on the gradient of the objective function to `1e-8`

.

CVMdl =
ClassificationPartitionedLinear
CrossValidatedModel: 'Linear'
ResponseName: 'Y'
NumObservations: 31572
KFold: 5
Partition: [1x1 cvpartition]
ClassNames: [0 1]
ScoreTransform: 'none'
Properties, Methods

`CVMdl`

is a `ClassificationPartitionedLinear`

model. Because `fitclinear`

implements 5-fold cross-validation, `CVMdl`

contains 5 `ClassificationLinear`

models that the software trains on each fold.

Estimate the *k*-fold margins for each regularization strength.

`m`

is a 31572-by-11 matrix of cross-validated margins for each observation. The columns correspond to the regularization strengths.

Plot the *k*-fold margins for each regularization strength. Because logistic regression scores are in [0,1], margins are in [-1,1]. Rescale the margins to help identify the regularization strength that maximizes the margins over the grid.

Several values of `Lambda`

yield *k*-fold margin distributions that are compacted near 10000. Higher values of lambda lead to predictor variable sparsity, which is a good quality of a classifier.

Choose the regularization strength that occurs just before the centers of the *k*-fold margin distributions start decreasing.

Train a linear classification model using the entire data set and specify the desired regularization strength.

To estimate labels for new observations, pass `MdlFinal`

and the new data to `predict`

.