Data science

Skill TheWatcher01/skills/.claude/skills/data-science

Agent Skills collection — Reusable capabilities for AI coding agents. Install: npx skills add TheWatcher01/skills

Install

npx -y skills add TheWatcher01/skills --skill data-science

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Agent data scientist spécialisé dans l'exploitation du datalake SubventionAI (2.8M entités, 168K versements, scoring ML, embeddings bge-m3 1024d). Stack: Python, PostgreSQL+pgvector, scikit-learn, XGBoost, LightGBM, spaCy, pandas. Use when: feature engineering, model training/evaluation, scoring batch, analyse statistique sectorielle, corrélation, clustering, prédiction, classification, anomaly detection, NLP sur objets sociaux, requêtes SQL analytiques, optimisation hyperparamètres. Triggers: data science, ML, modèle, scoring, feature engineering, prédiction, classification, clustering, analyse sectorielle, fundability, anomalie, NLP, embeddings, train, evaluate, cross-validation, AUC, F1, SHAP, Optuna.

SKILL.md

2.6 KB, 585 tokens by cl100k_base, as published. Nobody here has run it

Data Science Agent — Datalake SubventionAI

Données disponibles

Consulter references/schema.md pour le schéma complet.

Table	Rows	Clé
dl_entities	2.8M	SIREN, NAF, géo, scores, industryData JSONB
dl_subventions_versees	168K	Historique versements réels 2010-2026
dl_subventions	2413	Programmes actifs
dl_subvention_anomalies	3.1K	Anomalies détectées
dl_bodacc_annonces	-	Annonces légales
dl_marches_publics	-	Marchés publics

Connexion : PGPASSWORD=datalake_dev psql -h 127.0.0.1 -p 5434 -U datalake -d datalake

Workflow ML

Extraction : SQL paramétré, export CSV pour batch compute
Feature engineering : pandas + NumPy dans scripts/ml/
Training : scikit-learn/XGBoost/LightGBM, cross-validation 5-fold
Evaluation : AUC-ROC, F1, precision/recall, SHAP feature importance
Tuning : Optuna (TPE sampler, 100 trials)
Scoring batch : export → compute local → reimport PG (séparer compute et I/O)
Persistance : colonnes dédiées dans dl_entities

Scoring existant

Score	Colonne	Modèle
Fundability	fundabilityScore	Batch SQL 6 critères
Digital maturity	digitalMaturityScore	7 facteurs rule-based
CNIL risk	cnilRiskScore	XGBoost AUC 0.90
Hot lead	hotLeadScore	Composite 6 dimensions

Règles critiques

SQL toujours paramétré ($1, $2) — jamais d'interpolation
Données factuelles uniquement — jamais de chiffres approximatifs
Performance first : batch SQL, COPY pour imports, pas de N+1
Scripts dans scripts/ml/ avec docstring et if __name__ == "__main__":
Modèles évalués avec cross-validation avant déploiement
Embeddings : bge-m3 1024d, colonne vector(1024) pgvector, index HNSW

Patterns courants

Consulter references/patterns.md pour templates SQL et Python.

What ships with it: 2 files

5.5 KB alongside SKILL.md

references/

patterns.md2.9 KB
schema.md2.6 KB