Abstract
The problem of counterfactual explanations is that of minimally adjusting attributes in a source input instance so that it is classified as a target class under a given classifier. They answer practical questions of the type “what should my annual income be for my loan to be approved?”, for example. We focus on classification and regression trees, both axis-aligned and oblique (having hyperplane splits), and formulate the counterfactual explanation as an optimization problem. Although this problem is nonconvex and nondifferentiable, an exact solution can be computed very efficiently, even with high-dimensional feature vectors and with both continuous and categorical features. We also show how the counterfactual explanation formulation can answer a range of important practical questions, providing a way to query a trained tree and suggest possible actions to overturn its decision, and demonstrate it in several case studies. The results are particularly relevant for finance, medicine or legal applications, where interpretability and counterfactual explanations are particularly important.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsReferences
Bella, A., Ferri, C., Hernández-Orallo, J., Ramírez-Quintana, M.J.: Using negotiable features for prescription problems. Computing 91, 135–168 (2011)
Breiman, L.J., Friedman, J.H., Olshen, R.A., Stone, C.J.: Classification and Regression Trees. Wadsworth, Belmont (1984)
Carreira-Perpiñán, M.Á.: The Tree Alternating Optimization (TAO) algorithm: a new way to learn decision trees and tree-based models (2021). arXIV
Carreira-Perpiñán, M.Á., Hada, S.S.: Counterfactual explanations for oblique decision trees: exact, efficient algorithms. In: Proceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI 2021), 2–9 February 2021, pp. 6903–6911 (2021)
Carreira-Perpiñán, M.Á., Hada, S.S.: Counterfactual explanations for oblique decision trees: exact, efficient algorithms. arXiv:2103.01096 (2021)
Carreira-Perpiñán, M.Á., Hada, S.S.: Inverse classification with logistic and softmax classifiers: efficient optimization (2021). arXIV
Carreira-Perpiñán, M.Á., Tavallali, P.: Alternating optimization of decision trees, with application to learning sparse oblique trees. In: Bengio, S., Wallach, H., Larochelle, H., Grauman, K., Cesa-Bianchi, N., Garnett, R. (eds.) Advances in Neural Information Processing Systems (NEURIPS), vol. 31, pp. 1211–1221. MIT Press, Cambridge (2018)
Carreira-Perpiñán, M.Á., Zharmagambetov, A.: Ensembles of bagged TAO trees consistently improve over random forests, AdaBoost and gradient boosting. In: Proceedings of the 2020 ACM-IMS Foundations of Data Science Conference (FODS 2020), Seattle, WA, 19–20 October 2020, pp. 35–46(2020)
Cui, Z., Chen, W., He, Y., Chen, Y.: Optimal action extraction for random forests and boosted trees. In: Proceedings of the 21st ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (SIGKDD 2015), Sydney, Australia, 10–13 August 2015, pp. 179–188 (2015)
Dosovitskiy, A., Brox, T.: Inverting visual representations with convolutional networks. In: Proceedings of the 2016 IEEE Computer Society Conference Computer Vision and Pattern Recognition (CVPR’16), Las Vegas, NV, 26 June–1 July 2016 (2016)
Freitas, A.A.: Comprehensible classification models: a position paper. SIGKDD Explor. 15(1), 1–10 (2014)
Gabidolla, M., Zharmagambetov, A., Carreira-Perpiñán, M.Á.: Improved multiclass adaboost using sparse oblique decision trees (2021), submitted
Goodfellow, I.J., Shlens, J., Szegedy, C.: Explaining and harnessing adversarial examples. In: Proceedings of the 3rd International Conference on Learning Representations (ICLR 2015), San Diego, CA, 7–9 May 2015 (2015)
Guidotti, R., Monreale, A., Ruggieri, S., Turini, F., Giannotti, F., Pedreschi, D.: A survey of methods for explaining black box models. ACM Comput. Surv. 51(5), 93 (2018)
Gurobi Optimization, LLC: Gurobi optimizer reference manual (2019)
Hada, S.S., Carreira-Perpiñán, M.Á.: Sampling the “inverse set” of a neuron: an approach to understanding neural nets. arXiv:1910.04857 (2019)
Hada, S.S., Carreira-Perpiñán, M.Á.: Sampling the “inverse set” of a neuron. In: IEEE International Conference on Image Processing (ICIP 2021), Anchorage, AK, 9–12 September 2021 (2021)
Hada, S.S., Carreira-Perpiñán, M.Á., Zharmagambetov, A.: Sparse oblique decision trees: a tool to understand and manipulate neural net features. arXiv:2104.02922 (2021)
Hada, S.S., Carreira-Perpiñán, M.Á., Zharmagambetov, A.: Understanding and manipulating neural net features using sparse oblique classification trees. In: IEEE International Conference on Image Processing (ICIP 2021), Anchorage, AK, 19–12 September 2021 (2021)
Idelbayev, Y., Zharmagambetov, A., Gabidolla, M., Carreira-Perpiñán, M.Á.: Faster neural net inference via forests of sparse oblique decision trees (2021). arXIV
Lipton, Z.C.: The mythos of model interpretability. Comm. ACM 81(10), 36–43 (2018)
Mahendran, A., Vedaldi, A.: Visualizing deep convolutional neural networks using natural pre-images. Int. J. Comput. Vision 120(3), 233–255 (2016)
Martens, D., Provost, F.: Explaining data-driven document classifications. MIS Q. 38(1), 73–99 (2014)
Russell, C.: Efficient search for diverse coherent explanations. In: Proceedings of ACM Conference Fairness, Accountability, and Transparency (FAT 2019), Atlanta, GA, 29–31 January 2019, pp. 20–28 (2019)
Simonyan, K., Vedaldi, A., Zisserman, A.: Deep inside convolutional networks: visualising image classification models and saliency maps. In: Proceedings of the 2nd International Conference Learning Representations (ICLR 2014), Banff, Canada, 14–16 April 2014 (2014)
Szegedy, C., et al.: Intriguing properties of neural networks. In: Proceedings of the 2nd International Conference on Learning Representations (ICLR 2014), Banff, Canada, 14–16 April 2014 (2014)
Ustun, B., Spangher, A., Liu, Y.: Actionable recourse in linear classification. In: Proceedings of ACM Conference Fairness, Accountability, and Transparency (FAT 2019), Atlanta, GA, 29–31 January 2019, pp. 10–19 (2019)
Wachter, S., Mittelstadt, B., Russell, C.: Counterfactual explanations without opening the black box: automated decisions and the GDPR. Harvard J. Law Technol. 31(2), 841–887 (2018)
Wu, X., Kumar, V. (eds.): The Top Ten Algorithms in Data Mining. Chapman & Hall/CRC Data Mining and Knowledge Discovery Series, CRC Publishers, Boca Raton (2009)
Yang, Q., Yin, J., Ling, C.X., Pan, R.: Extracting actionable knowledge from decision trees. IEEE Trans. Knowl. Data Eng. 18(1), 43–56 (2006)
Zeiler, M.D., Fergus, R.: Visualizing and understanding convolutional networks. In: Proceedings of 13th European Conference Computer Vision (ECCV’14), Zürich, Switzerland, 6–12 September 2014, pp. 818–833 (2014)
Zhang, C., Liu, C., Zhang, X., Almpanidis, G.: An up-to-date comparison of state-of-the-art classification algorithms. Expert Syst. Appl. 82, 128–150 (2017)
Zharmagambetov, A., Carreira-Perpiñán, M.Á.: Smaller, more accurate regression forests using tree alternating optimization. In: Daumé III, H., Singh, A. (eds.) Proceedings of the 37th International Conference on Machine Learning (ICML 2020), 13–18 July 2020, pp. 11398–11408 (2020)
Zharmagambetov, A., Carreira-Perpiñán, M.Á.: Learning a tree of neural nets. In: Proceedings of the IEEE International Conference Acoustics, Speech and Signal Processing (ICASSP’21), , Toronto, Canada, 6–11 June 2021, pp. 3140–3144 (2021)
Zharmagambetov, A., Gabidolla, M., Carreira-Perpiñán, M.Á.: Improved boosted regression forests through non-greedy tree optimization. In: International Joint Conference Neural Networks (IJCNN’21), Virtual Event, 18–22 July 2021 (2021)
Zharmagambetov, A., Gabidolla, M., Carreira-Perpiñán, M.Á.: Improved multiclass AdaBoost for image classification: the role of tree optimization. In: IEEE International Conference on Image Processing (ICIP 2021), Anchorage, AK, 19–22 September 2021 (2021)
Zharmagambetov, A., Gabidolla, M., Carreira-Perpiñán, M.Á.: Softmax tree: an accurate, fast classifier when the number of classes is large (2021), submitted
Zharmagambetov, A., Hada, S.S., Gabidolla, M., Carreira-Perpiñán, M.Á.: Non-greedy algorithms for decision tree optimization: an experimental comparison. In: International Joint Conference on Neural Networks (IJCNN’21), Virtual event, 18–22 July 2021 (2021)
Acknowledgments
Work partially supported by NSF award IIS–2007147.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Hada, S.S., Carreira-Perpiñán, M.Á. (2021). Exploring Counterfactual Explanations for Classification and Regression Trees. In: Kamp, M., et al. Machine Learning and Principles and Practice of Knowledge Discovery in Databases. ECML PKDD 2021. Communications in Computer and Information Science, vol 1524. Springer, Cham. https://doi.org/10.1007/978-3-030-93736-2_37
Download citation
DOI: https://doi.org/10.1007/978-3-030-93736-2_37
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-93735-5
Online ISBN: 978-3-030-93736-2
eBook Packages: Computer ScienceComputer Science (R0)