.BT_relative_influence {BT}

Method for estimating the relative influence.

Description

Helper function for computing the relative influence of each variable in the BT object.

Usage

.BT_relative_influence(
  BTFit_object,
  n.iter,
  rescale = FALSE,
  sort.it = FALSE,
  consider.competing = FALSE,
  consider.surrogates = FALSE
)

Arguments

BTFit_object

a BTFit object.

n.iter

number of boosting iterations used for computation. If not provided, the function will perform a best guess approach to determine the optimal number of iterations. In fact, if a validation set was used during the fitting, the retained number of iterations is the one corresponding to the lowest validation set error ; otherwise, if cross-validation was performed, the number of iterations resulting in lowest cross-validation error will be used; otherwise, if the out-of-bag parameter was defined, the OOB error will be used to determine the optimal number of iterations; otherwise, all iterations will be used.

rescale

whether or not the results should be rescaled (divided by the maximum observation). Default set to FALSE.

sort.it

whether or not the results should be (reverse) sorted. Default set to FALSE.

consider.competing

whether or not competing split should be considered in the relative influence computation. Default set to FALSE.

consider.surrogates

whether or not surrogates should be considered in the relative influence computation. Default set to FALSE.

Details

This function is not intended for end-user use. It performs the relative influence computation and is called during the summary function. Note that a permutation approach is not yet implemented.

Value

Returns by default an unprocessed vector of estimated relative influences. If the rescale and sort.it arguments are used, it returns a processed version of the same vector.

Author(s)

Gireg Willame [e-mail]

This package is inspired by the gbm3 package. For more details, see https://github.com/gbm-developers/gbm3/.

References

M. Denuit, D. Hainaut and J. Trufin (2019). Effective Statistical Learning Methods for Actuaries |: GLMs and Extensions, Springer Actuarial.

M. Denuit, D. Hainaut and J. Trufin (2019). Effective Statistical Learning Methods for Actuaries ||: Tree-Based Methods and Extensions, Springer Actuarial.

M. Denuit, D. Hainaut and J. Trufin (2019). Effective Statistical Learning Methods for Actuaries |||: Neural Networks and Extensions, Springer Actuarial.

M. Denuit, D. Hainaut and J. Trufin (2022). Response versus gradient boosting trees, GLMs and neural networks under Tweedie loss and log-link. Accepted for publication in Scandinavian Actuarial Journal.

M. Denuit, J. Huyghe and J. Trufin (2022). Boosting cost-complexity pruned trees on Tweedie responses: The ABT machine for insurance ratemaking. Paper submitted for publication.

M. Denuit, J. Trufin and T. Verdebout (2022). Boosting on the responses with Tweedie loss functions. Paper submitted for publication.

See Also

BT, BTFit, BT_perf.