ECO-AIM-AI-014¶
Name: Inefficient feature preprocessing pipelines
Category: AIM
Family: AI
Primary layer: ai
System layers: ai
Description¶
Preprocessing waste increases training and inference cost.
Impact¶
- confidence: 0.55
- notes: Measure and optimize hotspots.
- type: cpu
Detection¶
- languages:
- python
- method: trace
Remediation¶
- guidance: Cache features; vectorize; batch transforms.
- tradeoffs: Complexity.
Pattern examples¶
No pattern examples provided.
Remediation examples¶
No remediation examples provided.