Skip to content

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"
  ]
}