ic_007 · Industry: Education — course-dropout prediction
Status: ✓ Passing · Last run: 2026-06-07T07:16:32 · Pipeline: v2
· History: 1/1 runs passed (100%)
What was tested
Education — course-dropout prediction Dropout prediction: students + courses + grades + attendance + labels. Signal: low attendance + declining grades.
What we planted in the data
- 300 entities
- Signal strength: 0.70 (sigmoid slope multiplier)
What the system did
- Training rows built: 600
- Features used in model: 18
- Model selected:
RandomForestClassifier - CV AUC: 0.773
Top features by importance:
score_pct_avg_14d(0.179)prior_gpa(0.178)score_pct_max_7d(0.106)score_pct_avg_all_time_sum(0.091)attendance_rate_60d(0.087)
What we checked — all assertions passed
| Status | Assertion | Detail |
|---|---|---|
| ✓ | no_pipeline_errors |
clean run |
| ✓ | feature_survival_rate_above |
18/29 features survived (62% ≥ 40%) |
| ✓ | source_tables_contributing_above |
5 distinct source-table prefixes in surviving features (['attendance', 'major', 'prior', 'score', 'year']) |
| ✓ | auc_in_band |
AUC = 0.773 inside band [0.55, 0.95] |
How this could have gone wrong
(no assertion descriptions provided)
Why this case matters
Industry-canonical ML problem; multi-table shape mirrors real customer datasets.
Reproducing
# from auto_insight_api/
python -m validation.v2 run ic_007 --pipeline v2 -v
- Case config:
validation/v2/cases/ic_007_education_dropout.yaml - Data shape:
industry_education_dropout - Analytics type:
ml_binary