Counterfactual explanations focus on “actionable knowledge” to help end-users understand how a machine learning outcome could be changed to one that is more desirable. For this purpose a counterfactual explainer needs to be able to reason with similarity knowledge in order to discover input dependencies that relate to outcome changes. Identifying the minimum subset of feature changes to action a change in the decision is an interesting challenge for counterfactual explainers. In this paper we show how feature relevance based explainers (such as LIME), can be combined with a counterfactual explainer to identify the minimum sub-set of “actionable features”. We demonstrate our hybrid approach on a real-world use case on student outcome prediction using data from the Campus Moodle Virtual Learning environment. Our preliminary results demonstrate that counterfactual feature weighting to be a viable strategy that should be adopted to minimise the number of actionable changes.
WIJEKOON, A., WIRATUNGA, N., NKILSI-ORJI, I., MARTIN, K., PALIHAWADANA, C. and CORSAR, D. 2021. Counterfactual explanations for student outcome prediction with Moodle footprints. In Martin, K., Wiratunga, N. and Wijekoon, A. (eds.) SICSA XAI workshop 2021: proceedings of 2021 SICSA (Scottish Informatics and Computer Science Alliance) eXplainable artificial intelligence workshop (SICSA XAI 2021), 1st June 2021, [virtual conference]. CEUR workshop proceedings, 2894. Aachen: CEUR-WS [online], session 1, pages 1-8. Available from: http://ceur-ws.org/Vol-2894/short1.pdf