{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2024,9,16]],"date-time":"2024-09-16T13:20:35Z","timestamp":1726492835572},"reference-count":37,"publisher":"Wiley","issue":"3","license":[{"start":{"date-parts":[[2024,6,9]],"date-time":"2024-06-09T00:00:00Z","timestamp":1717891200000},"content-version":"vor","delay-in-days":8,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"}],"funder":[{"DOI":"10.13039\/501100004564","name":"Ministarstvo Prosvete, Nauke i Tehnolo\u0161kog Razvoja","doi-asserted-by":"publisher","award":["451\u201003\u201047\/2023\u201001\/200358"],"id":[{"id":"10.13039\/501100004564","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Computational Intelligence"],"published-print":{"date-parts":[[2024,6]]},"abstract":"Abstract<\/jats:title>Cost\u2010sensitive ensemble learning as a combination of two approaches, ensemble learning and cost\u2010sensitive learning, enables generation of cost\u2010sensitive tree\u2010based ensemble models using the cost\u2010sensitive decision tree (CSDT) learning algorithm. In general, tree\u2010based models characterize nice graphical representation that can explain a model's decision\u2010making process. However, the depth of the tree and the number of base models in the ensemble can be a limiting factor in comprehending the model's decision for each sample. The CSDT models are widely used in finance (e.g., credit scoring and fraud detection) but lack effective explanation methods. We previously addressed this gap with cost\u2010sensitive tree Shapley Additive Explanation Method (CSTreeSHAP), a cost\u2010sensitive tree explanation method for the single\u2010tree CSDT model. Here, we extend the introduced methodology to cost\u2010sensitive ensemble models, particularly cost\u2010sensitive random forest models. The paper details the theoretical foundation and implementation details of CSTreeSHAP for both single CSDT and ensemble models. The usefulness of the proposed method is demonstrated by providing explanations for single and ensemble CSDT models trained on well\u2010known benchmark credit scoring datasets. Finally, we apply our methodology and analyze the stability of explanations for those models compared to the cost\u2010insensitive tree\u2010based models. Our analysis reveals statistically significant differences between SHAP values despite seemingly similar global feature importance plots of the models. This highlights the value of our methodology as a comprehensive tool for explaining CSDT models.<\/jats:p>","DOI":"10.1111\/coin.12651","type":"journal-article","created":{"date-parts":[[2024,6,10]],"date-time":"2024-06-10T02:48:47Z","timestamp":1717987727000},"update-policy":"http:\/\/dx.doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["Cost\u2010sensitive tree SHAP for explaining cost\u2010sensitive tree\u2010based models"],"prefix":"10.1111","volume":"40","author":[{"ORCID":"http:\/\/orcid.org\/0000-0003-1176-5677","authenticated-orcid":false,"given":"Marija","family":"Kopanja","sequence":"first","affiliation":[{"name":"Center for Information Technologies BioSense Institute Novi Sad Serbia"}]},{"ORCID":"http:\/\/orcid.org\/0000-0002-1067-6333","authenticated-orcid":false,"given":"Stefan","family":"Ha\u010dko","sequence":"additional","affiliation":[{"name":"Center for Information Technologies BioSense Institute Novi Sad Serbia"}]},{"ORCID":"http:\/\/orcid.org\/0000-0002-2259-4693","authenticated-orcid":false,"given":"Sanja","family":"Brdar","sequence":"additional","affiliation":[{"name":"Center for Information Technologies BioSense Institute Novi Sad Serbia"}]},{"ORCID":"http:\/\/orcid.org\/0000-0003-1267-5411","authenticated-orcid":false,"given":"Milo\u0161","family":"Savi\u0107","sequence":"additional","affiliation":[{"name":"Department for Mathematics and Informatics Faculty of Sciences Novi Sad Serbia"}]}],"member":"311","published-online":{"date-parts":[[2024,6,9]]},"reference":[{"key":"e_1_2_12_2_1","doi-asserted-by":"publisher","DOI":"10.1145\/502512.502540"},{"key":"e_1_2_12_3_1","first-page":"973","volume-title":"Proceedings of the 17th International Joint Conference on Artificial Intelligence","author":"Elkan C","year":"2001"},{"key":"e_1_2_12_4_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eswa.2016.03.045"},{"key":"e_1_2_12_5_1","series-title":"CEUR Workshop Proceedings","first-page":"95","volume-title":"Joint Proceedings of the xAI\u20102023 Late\u2010Breaking Work, Demos and Doctoral Consortium Co\u2010Located with the 1st World Conference on eXplainable Artificial Intelligence (xAI\u20102023), Lisbon, Portugal, July 26\u201028, 2023","author":"Kopanja M","year":"2023"},{"key":"e_1_2_12_6_1","unstructured":"BahnsenAC AouadaD OtterstenBE.Ensemble of example\u2010dependent cost\u2010sensitive decision trees.ArXiv abs\/1505.04637.2015."},{"key":"e_1_2_12_7_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-1-349-20181-5_25"},{"key":"e_1_2_12_8_1","doi-asserted-by":"publisher","DOI":"10.1038\/s42256-019-0138-9"},{"key":"e_1_2_12_9_1","unstructured":"MitchellR FrankE HolmesG.Gputreeshap: massively parallel exact calculation of shap scores for tree ensembles.2020. doi:10.48550\/ARXIV.2010.13972"},{"key":"e_1_2_12_10_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eswa.2015.04.042"},{"key":"e_1_2_12_11_1","doi-asserted-by":"publisher","DOI":"10.1109\/34.310684"},{"key":"e_1_2_12_12_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICDM.2003.1250950"},{"key":"e_1_2_12_13_1","first-page":"263","volume-title":"2014 13th International Conference on Machine Learning and Applications","author":"Correa BA","year":"2014"},{"key":"e_1_2_12_14_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICMLA.2013.68"},{"key":"e_1_2_12_15_1","doi-asserted-by":"publisher","DOI":"10.1007\/1187560415"},{"key":"e_1_2_12_16_1","doi-asserted-by":"publisher","DOI":"10.1109\/CEC.2005.1554956"},{"key":"e_1_2_12_17_1","doi-asserted-by":"publisher","DOI":"10.1145\/1015330.1015369"},{"key":"e_1_2_12_18_1","doi-asserted-by":"publisher","DOI":"10.1109\/TKDE.2002.1000348"},{"key":"e_1_2_12_19_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eswa.2013.05.021"},{"key":"e_1_2_12_20_1","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2010.71"},{"key":"e_1_2_12_21_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.elerap.2017.06.004"},{"key":"e_1_2_12_22_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICMLA55696.2022.00225"},{"key":"e_1_2_12_23_1","doi-asserted-by":"crossref","DOI":"10.1186\/s12911-021-01436-7","article-title":"A hybrid cost\u2010sensitive ensemble for heart disease prediction","volume":"21","author":"Zhenya Q","year":"2021","journal-title":"BMC Med Inform Decis Mak"},{"key":"e_1_2_12_24_1","doi-asserted-by":"publisher","DOI":"10.1007\/s10618-021-00790-4"},{"key":"e_1_2_12_25_1","unstructured":"Costcla \u2010 a python module for cost\u2010sensitive machine learning.https:\/\/pypi.org\/project\/costcla\/Accessed December 19 2022."},{"key":"e_1_2_12_26_1","unstructured":"DablainD BellingerC AhaW ChawlaN KrawczykB.understanding imbalanced data: xAI & interpretable ML framework.ResarchGate Preprint.2022."},{"key":"e_1_2_12_27_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.infsof.2022.107016"},{"key":"e_1_2_12_28_1","doi-asserted-by":"publisher","DOI":"10.1007\/978\u20101\u20100716\u20101418\u20101"},{"key":"e_1_2_12_29_1","doi-asserted-by":"publisher","DOI":"10.1109\/34.709601"},{"key":"e_1_2_12_30_1","first-page":"4765","volume-title":"Advances in Neural Information Processing Systems","author":"Lundberg S","year":"2017"},{"key":"e_1_2_12_31_1","first-page":"2825","article-title":"Scikit\u2010learn: machine learning in python","volume":"12","author":"Pedregosa F","year":"2011","journal-title":"J Mach Learn Res"},{"key":"e_1_2_12_32_1","unstructured":"W. C. Credit Fusion.Give me some credit.2011https:\/\/kaggle.com\/competitions\/GiveMeSomeCredit"},{"key":"e_1_2_12_33_1","doi-asserted-by":"publisher","DOI":"10.3233\/AF-180227"},{"key":"e_1_2_12_34_1","doi-asserted-by":"publisher","DOI":"10.1109\/ISCIPT53667.2021.00016"},{"key":"e_1_2_12_35_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.jfds.2022.07.002"},{"key":"e_1_2_12_36_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eswa.2022.118909"},{"key":"e_1_2_12_37_1","doi-asserted-by":"crossref","unstructured":"CorreaBA StojanovicA AouadaD OtterstenB.Improving credit card fraud detection with calibrated probabilities. Paper presented at: Proceedings of the 2014 SIAM International Conference on Data Mining.2014; Pennsylvania PA. doi:10.1137\/1.9781611973440.78","DOI":"10.1137\/1.9781611973440.78"},{"key":"e_1_2_12_38_1","doi-asserted-by":"publisher","DOI":"10.1214\/aoms\/1177730491"}],"container-title":["Computational Intelligence"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/coin.12651","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,7,16]],"date-time":"2024-07-16T00:44:16Z","timestamp":1721090656000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1111\/coin.12651"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,6]]},"references-count":37,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2024,6]]}},"alternative-id":["10.1111\/coin.12651"],"URL":"https:\/\/doi.org\/10.1111\/coin.12651","archive":["Portico"],"relation":{},"ISSN":["0824-7935","1467-8640"],"issn-type":[{"value":"0824-7935","type":"print"},{"value":"1467-8640","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,6]]},"assertion":[{"value":"2023-12-29","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2024-04-10","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2024-06-09","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}