An integrated explainable AI framework with counterfactual optimization for decision support

Nusrat Jahan, Jannatul Maua, Uland Rozario, Hashibul Ahsan Shoaib · 2026

Read An integrated explainable AI framework with counterfactual optimization for decision support by Nusrat Jahan, Jannatul Maua, Uland Rozario, Hashibul Ahsan Shoaib (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0358643.txt.

Description

Nusrat Jahan and three colleagues build a decision support system that explains its own predictions and suggests what to change, in PLOS ONE in 2026. Managers like the accuracy of deep learning but distrust answers they cannot inspect. The framework combines a predictor with feature attribution, which shows what drove each call, and counterfactual reasoning, which finds the smallest changes that would flip an outcome.

They tested it on employee attrition and customer churn data, separately and combined. Accuracy reached roughly ninety-one percent for attrition and eighty-seven percent for churn, with well calibrated confidence scores. The explanations scored high on fidelity checks, and the suggested actions cost less than those from comparison methods. For a manager, that means not just a warning that someone may leave, but a plausible, affordable lever to pull first.

Catalog record

CREATORNusrat Jahan, Jannatul Maua, Uland Rozario, Hashibul Ahsan Shoaib
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE82426
LICENSECC_BY
RIGHTSCC BY
SOURCEPLOS

Source: https://doi.org/10.1371/journal.pone.0358643

Browse the full library · About the Tower of Babel