Generate prescriptive feedback via counterfactual modelling constrained to actionable features¶

Strategy · All strategies
Evidence · no claims cited

Description¶

For students predicted high-risk, the dashboard models a minimal set of changes to the student's input values that would flip the prediction to low risk, using counterfactual modelling. Changes are constrained "to only the features that are actionable, thus ignoring immutable features like citizenship." Outputs are converted into human-understandable textual feedback offering data-driven personalized suggestions for modifying learning behaviours.

Design Implications¶

Context¶

Requirements¶

• A trained predictive model and student feature data must be available; counterfactuals must be restricted to actionable (modifiable) features

Constraints¶

• Counterfactual outputs depend on the predictive model's accuracy and may be undermined by erroneous predictions, which the article notes can erode trust

Target Learners¶

• tertiary students identified as at risk of course non-completion

Target Learning Goals¶

• adjusting learning behaviour
• improving course outcomes
• self-regulated learning

Affordances¶

• Lad Theoretical Grounding Sct Srl Tl

Related Strategies¶

• Sensenablr Dashboard

Examples¶

-

Key Sources¶

• Gomathy Ramaswami, Teo Susnjak and Anuradha Mathrani. (2023). Effectiveness of a Learning Analytics Dashboard for Increasing Student Engagement Levels. Journal of Learning Analytics, 10(3). https://doi.org/10.18608/jla.2023.7935