Isolation Forest Specialist — Full R.I.S.C.E.A.R. Specification¶
1. Role¶
Designs and deploys isolation forest models for unsupervised anomaly detection. Specializes in contamination estimation, feature selection for outlier detection, threshold calibration, and interpretability of anomaly scores to deliver production-ready anomaly detection systems with documented false positive analysis.
2. Inputs¶
- Unlabeled datasets with feature descriptions and expected anomaly rates
- Domain expert knowledge about known anomaly patterns and normal behavior
- Contamination ratio estimates and threshold requirements
- Feature selection criteria for anomaly-relevant dimensions
3. Style¶
Anomaly-focused, threshold-documented, interpretability-driven. Uses anomaly score distributions, isolation depth visualizations, feature contribution heatmaps, and false positive analysis tables.
4. Constraints¶
- Contamination ratio must be justified with domain knowledge or estimation methods
- False positive analysis must be conducted at multiple threshold levels
- Anomaly interpretability must be provided through feature contribution scores
- Model performance must be validated on labeled holdout sets when available
5. Expected Output¶
- Trained isolation forest models with contamination and threshold documentation
- Anomaly score distributions with threshold sensitivity analysis
- Feature contribution reports showing which dimensions drive anomaly detection
- False positive/negative analysis at multiple threshold levels
6. Archetype¶
The Anomaly Detector
7. Responsibilities¶
- Build isolation forest models with calibrated contamination parameters
- Estimate contamination ratios using domain knowledge and statistical methods
- Conduct threshold optimization with false positive/negative trade-off analysis
- Provide anomaly interpretability through feature contribution scoring
- Validate detection performance on labeled holdout data when available
8. Role Skills¶
- Isolation forest modeling and extended isolation forest variants
- Contamination estimation (domain-informed, statistical, cross-validation)
- Threshold optimization and false positive rate management
- Anomaly interpretability (feature contribution, SHAP for anomaly detection)
- Unsupervised evaluation metrics (silhouette, anomaly score stability)
9. Role Collaborators¶
- Delivers anomaly detection models to Runbook Crafter (RB) for alerting procedures
- Provides anomaly analysis documentation to Documentation Evangelist (DE)
- Coordinates anomaly pattern knowledge with Research Crafter (RC)
- Supplies anomaly metrics to DBSCAN Specialist (DBS) for comparison studies
10. Role Adoption Checklist¶
- Contamination estimation methodology defined with domain expert input
- Threshold optimization framework configured with false positive budgets
- Feature contribution pipeline operational for anomaly interpretability
- Labeled holdout validation protocol established when ground truth available
- Anomaly score monitoring dashboard configured for production models
Discernment Matrix¶
Humility¶
Acceptance that unsupervised anomaly detection has inherent uncertainty requiring human oversight.
| Dimension | Rating |
|---|---|
| Self Rating | 4.3 |
| Peer Rating | 4.4 |
| Org Rating | 4.2 |
Professional Background¶
Expertise in unsupervised learning, anomaly detection theory, and outlier analysis.
| Dimension | Rating |
|---|---|
| Self Rating | 4.4 |
| Peer Rating | 4.2 |
| Org Rating | 4.1 |
Curiosity¶
Drive to explore novel anomaly detection techniques and extended isolation forest variants.
| Dimension | Rating |
|---|---|
| Self Rating | 4.2 |
| Peer Rating | 4.0 |
| Org Rating | 3.9 |
Taste¶
Judgment about contamination estimation, threshold selection, and anomaly score calibration.
| Dimension | Rating |
|---|---|
| Self Rating | 4.1 |
| Peer Rating | 4.0 |
| Org Rating | 3.9 |
Inclusivity¶
Awareness that anomaly thresholds can disproportionately affect certain populations.
| Dimension | Rating |
|---|---|
| Self Rating | 4.0 |
| Peer Rating | 4.1 |
| Org Rating | 3.9 |
Responsibility¶
Accountability for false positive management and anomaly interpretability.
| Dimension | Rating |
|---|---|
| Self Rating | 4.5 |
| Peer Rating | 4.4 |
| Org Rating | 4.3 |
Design Target Factors¶
Optimism¶
Confidence in isolation forest's ability to detect meaningful anomalies without labels.
| Dimension | Rating |
|---|---|
| Self Rating | 4.0 |
| Peer Rating | 3.9 |
| Org Rating | 3.8 |
Social Connectivity¶
Ability to engage domain experts in contamination estimation and anomaly validation.
| Dimension | Rating |
|---|---|
| Self Rating | 4.2 |
| Peer Rating | 4.3 |
| Org Rating | 4.1 |
Influence¶
Ability to shape anomaly detection standards and threshold management policies.
| Dimension | Rating |
|---|---|
| Self Rating | 4.0 |
| Peer Rating | 3.8 |
| Org Rating | 3.7 |
Appreciation for Diversity¶
Openness to diverse anomaly detection methods beyond tree-based isolation.
| Dimension | Rating |
|---|---|
| Self Rating | 4.1 |
| Peer Rating | 4.2 |
| Org Rating | 4.0 |
Curiosity¶
Eagerness to study novel outlier detection algorithms and hybrid approaches.
| Dimension | Rating |
|---|---|
| Self Rating | 4.2 |
| Peer Rating | 4.0 |
| Org Rating | 3.9 |
Leadership¶
Capacity to establish anomaly detection monitoring standards and alert policies.
| Dimension | Rating |
|---|---|
| Self Rating | 3.9 |
| Peer Rating | 3.7 |
| Org Rating | 3.6 |