@article{fdi:010086049,
  title = {{T}he accuracy versus interpretability trade-off in fraud detection model},
  author = {{N}esvijevskaia, {A}. and {O}uillade, {S}. and {G}uilmin, {P}. and {Z}ucker, {J}ean-{D}aniel},
  editor = {},
  language = {{ENG}},
  abstract = {{L}ike a hydra, fraudsters adapt and circumvent increasingly sophisticated barriers erected by public or private institutions. {A}mong these institutions, banks must quickly take measures to avoid losses while guaranteeing the satisfaction of law-abiding customers. {F}acing an expanding flow of operations, effective banking relies on data analytics to support established risk control processes, but also on a better understanding of the underlying fraud mechanism. {I}n addition, fraud being a criminal offence, the evidential aspect of the process must also be considered. {T}hese legal, operational, and strategic constraints lead to compromises on the means to be implemented for fraud management. {T}his paper first focuses on the translation of practical questions raised in the banking industry at each step of the fraud management process into performance evaluation required to design a fraud detection model. {S}econdly, it considers a range of machine learning approaches that address these specificities: the imbalance between fraudulent and nonfraudulent operations, the lack of fully trusted labels, the concept-drift phenomenon, and the unavoidable trade-off between accuracy and interpretability of detection. {T}his state-of-the-art review sheds some light on a technology race between black box machine learning models improved by post-hoc interpretation and intrinsic interpretable models boosted to gain accuracy. {F}inally, it discusses how concrete and promising hybrid approaches can provide pragmatic, short-term answers to banks and policy makers without swallowing up stakeholders with economical and ethical stakes in this technological race.},
  keywords = {data analysis ; fraud detection ; human data mediation ; interpretability ; unbalanced data},
  booktitle = {},
  journal = {{D}ata and {P}olicy},
  volume = {3},
  numero = {},
  pages = {e12 [24 p.]},
  year = {2021},
  DOI = {10.1017/dap.2021.3},
  URL = {https://www.documentation.ird.fr/hor/fdi:010086049},
}