LSVM Build Options

Model Settings

Include intercept. Including the intercept (the constant term in the model) can increase the overall accuracy of the solution. If you can assume the data passes through the origin, you can exclude the intercept.

Sorting order for categorical target. Specifies the sorting order for the categorical target. This setting is ignored for continuous targets.

Regression precision (epsilon). Used only if the measurement level of the target field is Continuous. Causes errors to be accepted provided that they are less than the value specified here. Increasing the value may result in faster modeling, but at the expense of accuracy.

Exclude records with any missing values. When set to True, a record is excluded if any single value is missing.

Penalty Settings

Penalty function. Specifies the type of penalty function that is used to reduce the likelihood of overfitting. The options are either L1 or L2.

L1 and L2 reduce the chance of overfitting by adding a penalty on the coefficients. The difference between them is that when there is a large number of features, L1 uses feature selection by setting some coefficients to 0 during model building. L2 does not have this ability so should not be used when you have a large number of features.

Penalty parameter (lambda). Specifies the penalty (regularization) parameter. This setting is enabled if the Penalty function is set.