Package: IBLM 2.0.1

Karol Gawlowski

IBLM: Interpretable Boosted Linear Models

Implements Interpretable Boosted Linear Models (IBLMs). These combine a conventional generalized linear model (GLM) with a machine learning component, such as XGBoost. The package also provides tools within for explaining and analyzing these models. For more details see Gawlowski and Wang (2025) <https://ifoa-adswp.github.io/IBLM/reference/figures/iblm_paper.pdf>.

Authors:Karol Gawlowski [aut, cre, cph], Paul Beard [aut]

IBLM_2.0.1.tar.gz
IBLM_2.0.1.zip(r-4.7)IBLM_2.0.1.zip(r-4.6)IBLM_2.0.1.zip(r-4.5)
IBLM_2.0.1.tgz(r-4.6-any)IBLM_2.0.1.tgz(r-4.5-any)
IBLM_2.0.1.tar.gz(r-4.7-any)IBLM_2.0.1.tar.gz(r-4.6-any)
IBLM_2.0.1.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
IBLM/json (API)

# Install 'IBLM' in R:

install.packages('IBLM', repos = c('https://ifoa-adswp.r-universe.dev', 'https://cloud.r-project.org'))

Bug tracker:https://github.com/ifoa-adswp/iblm/issues

Pkgdown/docs site:https://ifoa-adswp.github.io

Datasets:

freMTPLmini - French Motor Insurance Claims Dataset

On CRAN:

4.93 score 1 stars 7 scripts 243 downloads 23 exports 69 dependencies

Last updated from:d495d193fe. Checks:9 OK. Indexed: yes.

Target	Result	Time
linux-devel-x86_64	OK	168
source / vignettes	OK	263
linux-release-x86_64	OK	191
macos-release-arm64	OK	234
macos-oldrel-arm64	OK	216
windows-devel	OK	137
windows-release	OK	135
windows-oldrel	OK	134
wasm-release	OK	129

Exports:beta_corrected_density beta_corrected_scatter beta_corrections_derive bias_density check_iblm_model correction_corridor create_beta_corrected_density create_beta_corrected_scatter create_bias_density create_overall_correction data_beta_coeff_booster data_beta_coeff_glm data_to_onehot explain_iblm extract_booster_shap get_pinball_scores load_freMTPL2freq overall_correction shap_to_onehot split_into_train_validate_test theme_iblm train_iblm_xgb train_xgb_as_per_iblm

Dependencies:base64enc bslib cachem cli colourpicker commonmark cpp11 data.table digest dplyr evaluate farver fastDummies fastmap fontawesome fs generics ggExtra ggplot2 glue gtable highr htmltools htmlwidgets httpuv isoband jquerylib jsonlite knitr labeling later lattice lifecycle magrittr Matrix memoise mime miniUI otel pillar pkgconfig promises purrr R6 rappdirs RColorBrewer Rcpp rlang rmarkdown S7 sass scales shiny shinyjs sourcetools statmod stringi stringr tibble tidyselect tinytex utf8 vctrs viridisLite withr xfun xgboost xtable yaml

IBLM

Rendered fromIBLM.Rmdusingknitr::rmarkdown

Last update: 2026-05-18
Started: 2025-11-23

Citation

Development and contributors

Readme and manuals

Help Manual

Help page	Topics
Density Plot of Beta Corrections for a Variable	beta_corrected_density
Scatter Plot of Beta Corrections for a Variable	beta_corrected_scatter
Compute Beta Corrections based on SHAP values	beta_corrections_derive
Density Plot of Bias Corrections from SHAP values	bias_density
Check Object of Class `iblm`	check_iblm_model
Plot GLM vs IBLM Predictions with Different Corridors	correction_corridor
Create Pre-Configured Beta Corrected Density Plot Function	create_beta_corrected_density
Create Pre-Configured Beta Corrected Scatter Plot Function	create_beta_corrected_scatter
Create Pre-Configured Bias Density Plot Function	create_bias_density
Create Pre-Configured Overall Correction Plot Function	create_overall_correction
Obtain Booster Model Beta Corrections for tabular data	data_beta_coeff_booster
Obtain GLM Beta Coefficients for tabular data	data_beta_coeff_glm
Convert Data Frame to Wide One-Hot Encoded Format	data_to_onehot
Explain GLM Model Predictions Using SHAP Values	explain_iblm
Extract SHAP values from an xgboost Booster model	extract_booster_shap extract_booster_shap.xgb.Booster
French Motor Insurance Claims Dataset	freMTPLmini
Calculate Pinball Scores for IBLM and Additional Models	get_pinball_scores
Load French Motor Third-Party Liability Frequency Dataset	load_freMTPL2freq
Plot Overall Corrections from Booster Component	overall_correction
Predict Method for IBLM	predict.iblm
Convert Shap values to Wide One-Hot Encoded Format	shap_to_onehot
Split Dataframe into: 'train', 'validate', 'test'	split_into_train_validate_test
Custom ggplot2 Theme for IBLM	theme_iblm
Train IBLM Model on XGBoost	train_iblm_xgb
Train XGBoost Model Using the IBLM Model Parameters	train_xgb_as_per_iblm