datascience

name	datascience
description	CSV analysis via 2-pass SOTA orchestrator (score 54/55 on 7 CSVs). Statistics, data profiling, visualizations, insights.
trigger_patterns	["analyse csv","analyser fichier",".csv","data analysis","statistiques"]
tools_required	["exec","pandas","llm_qwen35"]
examples	["/chemin/vers/fichier.csv","/tmp/patients.csv","~/data/sales_2026.csv"]
model	qwen3.5-35b-a3b
execution	{"command":"/home/remondiere/.openclaw/bin/analyze_csv.sh '$args'","timeout_ms":360000,"arg_mode":"raw"}

datascience — CSV Analysis

Dispatch direct vers analyze_csv.sh : orchestrateur 2-pass, score 54/55 sur 7 CSVs de référence.

Bypasse le modèle principal : le gateway appelle exec directement.

Process

1. Pass 1 — Data profiling (dtypes, null counts, unique values, distributions)
2. Pass 2 — LLM insights (patterns, anomalies, recommandations analytiques)

Output

• Shape et dtypes du dataset
• Statistiques descriptives (mean, std, min, max, quartiles)
• Null analysis + outliers
• Cardinalité + corrélations
• Insights LLM synthétisés en français