CostSensitiveLearning
The repository contains the R packages cslogit, csboost and PerformanceMetrics and the corresponding article:
Höppner, S., Baesens, B., Verbeke, W., and Verdonck, T. (2020). Instance- dependent cost-sensitive learning for detecting transfer fraud. arXiv:2005.02488
https://arxiv.org/abs/2005.02488
The R packages cslogit and csboost respectively contain the implementation of the cslogit and csboost algorithm for instance-dependent cost-sensitive learning using lasso-regularized logistic regression and gradient boosted decision trees, respectively. The R package PerformanceMetrics contains functions for performance measurement and threshold tuning for binary classification.
The R packages can be installed through the devtools package:
devtools::install_github("SebastiaanHoppner/CostSensitiveLearning/cslogit")
devtools::install_github("SebastiaanHoppner/CostSensitiveLearning/csboost")
devtools::install_github("SebastiaanHoppner/CostSensitiveLearning/PerformanceMetrics")