bestBoostingIter {logicDT}

Get the best number of boosting iterations

Description

This function can be used to compute the ideal number of boosting iterations for the fitted logic.boosted model using independent validation data.

Usage

bestBoostingIter(model, X, y, Z = NULL, consec.iter = 5, scoring_rule = "auc")

Arguments

model

Fitted logic.boosted model

X

Matrix or data frame of binary validation input data. This object should correspond to the binary matrix for fitting the model.

y

Validation response vector. 0-1 coding for binary outcomes.

Z

Optional quantitative covariables supplied as a matrix or data frame. Only used (and required) if the model was fitted using them.

consec.iter

Number of consecutive boosting iterations that do not increase the validation performance for determining the ideal number of iterations

scoring_rule

Scoring rule computing the validation performance. This can either be "auc" for the area under the receiver operating characteristic curve (default for binary reponses), "deviance" for the deviance, "nce" for the normalized cross entropy or "brier" for the Brier score. For regression purposes, the MSE (mean squared error) is automatically chosen.

Details

If the model performance (on the validation data) cannot be increased for consec.iter consecutive boosting iterations, the last iteration which increased the validation performance induces the ideal number of boosting iterations.

Value

The ideal number of boosting iterations