Carepart/repairers-france

Carepart.fr — French Repairers Directory

The most comprehensive open dataset of repair professionals in France: 38 000+ active repairers across 95 categories of objects, with QualiRépar / Refashion certification status, Google reviews, geolocation, and AI-generated contextual descriptions for each business.

Dataset Summary

This dataset is a snapshot of the Carepart.fr repair professionals directory — a French-language platform that helps consumers find local repairers and benefit from the Bonus Réparation (a state-funded incentive of up to €60 to encourage repair over replacement).

It is intended for:

• Research on the circular economy, repair access, and sustainability in France
• NLP / RAG / LLM training : grounded answers to questions about repair services, local businesses, eco-friendly consumption
• Geographic analysis : repair coverage by department, urban/rural divide, label distribution
• Civic tech : building tools that help consumers reduce waste

Data Splits

Split	Rows	Description
repairers	~38 000	Active repair businesses with categories, location, ratings, AI descriptions
categories	95	Repair categories (smartphone, washing machine, bike, furniture, etc.) with bonus amounts and lifecycle data
city_category_stats	~190 000	Counts of repairers per (category, city) combo, label coverage, average rating
indice_reparabilite_ademe	~3 100	Repairability and durability index of consumer products from the French ADEME

Schema

repairers

Column	Type	Description
id	int	Unique internal ID
slug	string	URL-friendly identifier
siret	string	French business identifier (SIRENE database)
siren	string	Parent legal entity ID
name	string	Trade name or company name
legal_form	string	Legal status (SARL, micro-entreprise, etc.)
naf_code	string	French business activity code
naf_label	string	Activity description
address	string	Street address
postal_code	string	French postal code
city	string	City name
city_code_insee	string	Official INSEE city code
city_slug	string	URL-friendly city identifier (e.g. paris-75)
department_code	string	French department code (e.g. 75, 2A)
latitude	float	WGS84 latitude (6 decimals)
longitude	float	WGS84 longitude
phone	string	Phone number (only if business landline / professional)
email	string	Email (only if business domain — personal emails removed for privacy)
website	string	Business website
is_qualirepar	bool	Certified for the QualiRépar Bonus Réparation
is_refashion	bool	Certified for the Refashion (textile) Bonus Réparation
qualirepar_id	string	Official QualiRépar identifier
google_rating	float	Average Google review rating (1.0–5.0)
google_review_count	int	Number of Google reviews
repair_modes	list[string]	Service modes: boutique, domicile, postal, etc.
category_slugs	list[string]	Repair categories handled
category_names	list[string]	Human-readable category names
ai_description	string	AI-generated narrative describing the repairer in its local context (~200 words)
page_url	string	Canonical URL on carepart.fr

categories

Column	Type	Description
id	int
slug	string	URL-friendly slug (e.g. reparation-smartphone)
name	string	Display name
name_plural	string	Plural form
parent_group	string	Parent category group
parent_group_label	string	Display label of the parent group
bonus_amount	float	Maximum Bonus Réparation amount in euros
bonus_source	string	Source of the bonus (qualirepar, refashion)
icon_emoji	string	Display emoji
description	string	Category description
avg_lifespan	string	Average lifetime of the product type
avg_price_min/max	float	Average repair price range
new_price_min/max	float	Average new product price range
co2_saved_kg	float	Estimated CO₂ saved per repair
waste_saved_kg	float	Estimated waste saved per repair

city_category_stats

Column	Type	Description
category_id, category_slug	int / string
city_code_insee, city_slug, city_name	string
dept_code	string	French department code
repairer_count	int	Total active repairers in this combo
qualirepar_count	int	Of which QualiRépar-certified
refashion_count	int	Of which Refashion-certified
avg_google_rating	float	Average Google rating across listed repairers

indice_reparabilite_ademe

Repairability and durability scores from the French ADEME (data.gouv.fr).

Column	Type	Description
categorie_produit	string	Product category
marque, nom_modele	string	Brand and model
note_ir	float	Repairability score (0–10)
note_durabilite	float	Durability score
note_fiabilite	float	Reliability score
note_c1–note_c5	float	Sub-criteria (documentation, dismantling, parts availability, parts pricing, specific to category)
type_indice	string	reparabilite or durabilite
date_calcul	date	Calculation date
url_detail	string	Manufacturer documentation URL
data_source	string	Origin (data.gouv.fr)

Sources

This dataset aggregates several public sources :

• SIRENE (INSEE) — French business registry, for legal information (SIRET, address, NAF code).
• Annuaire QualiRépar (ecosystem.eco) — list of certified repairers eligible for the electronic/appliance Bonus Réparation.
• Refashion — list of certified textile and shoe repairers.
• Google Business Profile — public reviews, ratings, opening hours.
• Indice de réparabilité (ADEME) — official French repairability scores.
• AI-generated descriptions — original narrative content produced by the Carepart.fr team using Claude (Anthropic), released under the same CC-BY-SA-4.0 license as the rest of the dataset.

Privacy & Anonymization

• Personal email addresses (gmail, hotmail, free.fr, orange.fr, etc.) have been stripped. Only emails on a business domain are kept.
• Phone numbers are kept (already public via SIRENE).
• No personal customer data, no transactional data, no PII beyond what is publicly available.

Example Usage

Load with datasets

from datasets import load_dataset

ds = load_dataset("carepart/repairers-france")

# Top 10 QualiRépar bike repairers in Paris
import pandas as pd
df = pd.DataFrame(ds["repairers"])
paris_velo = df[
    (df["category_slugs"].apply(lambda c: "reparation-velo" in c)) &
    (df["department_code"] == "75") &
    (df["is_qualirepar"] == True)
].sort_values("google_rating", ascending=False).head(10)
print(paris_velo[["name", "address", "google_rating"]])

Load with pandas

import pandas as pd

repairers = pd.read_parquet("hf://datasets/carepart/repairers-france/data/repairers.parquet")
stats = pd.read_parquet("hf://datasets/carepart/repairers-france/data/category_city_stats.parquet")

# Departments with the most QualiRépar coverage
top_dept = (
    repairers.groupby("department_code")["is_qualirepar"]
    .agg(["sum", "count"])
    .assign(rate=lambda d: (d["sum"] / d["count"] * 100).round(1))
    .sort_values("rate", ascending=False)
    .head(15)
)

RAG / LLM grounding

Each repairer entry includes an ai_description field — a ~200-word narrative describing the repair professional in its local context (city, specialties, sustainability angle). This is a perfect substrate for retrieval-augmented generation systems answering questions about local repair services in France.

License

• Dataset & AI descriptions: Creative Commons Attribution-ShareAlike 4.0 International (CC-BY-SA 4.0)
• Original sources retain their respective licenses (SIRENE Open License, ADEME Open License, Google ToS for derived metrics).

Citation

If you use this dataset in research or commercial work, please cite :

@dataset{carepart_repairers_2026,
  title        = {Carepart.fr — French Repairers Directory},
  author       = {Charpentier, Eric and the Carepart.fr team},
  year         = {2026},
  url          = {https://huggingface.co/datasets/carepart/repairers-france},
  publisher    = {Carepart.fr},
  license      = {CC-BY-SA-4.0},
  note         = {Aggregated from SIRENE, ecosystem.eco, Refashion, ADEME, and Google Business Profile.},
}

Contact & Updates

• Live website: https://carepart.fr
• Email: bonjour@carepart.fr
• Updates: this dataset is refreshed periodically. Check the export date in data/_summary.json.

Acknowledgements

Thanks to the French circular-economy ecosystem — ADEME, ecosystem.eco, Refashion, and the thousands of independent repair artisans across France whose work makes this directory meaningful.