ECO-AIM-AI-014 — Inefficient feature preprocessing pipelines¶
- Category: AI/ML (AIM)
- Family: AI (AI)
- Layer: AI
- Tier: 2
- Severity: note
- Tags: ai, pipelines
- Legacy ID: ECO-AI-014
Summary¶
Preprocessing waste increases training and inference cost.
Rationale¶
Preprocessing is often a hidden cost multiplier.
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."
}
Ontology¶
{
"system_layers": [
"ai"
]
}