grichardomi/restaurant-prices

MarketPulse Restaurant & Menu Prices — U.S. Metro Areas

Restaurant prices, menu prices, and food price benchmarking data for U.S. metros. This dataset contains 10,000 real-world menu item price observations from restaurant websites across 128 U.S. metro areas, with business location, normalized item names, and crawl metadata. It is designed for restaurant price comparison, menu price benchmarking, food inflation analysis, RAG, and model evaluation. Rolling 180-day window, updated regularly from the MarketPulse price index.

Why this dataset is different

Most public restaurant datasets are aggregated, synthetic, or limited to one city or chain. This one is:

• Observation-level rows — one row per menu item price, not averages or summaries
• Multi-metro with coordinates — 128+ U.S. metros with lat/lon per business
• Normalized item names — item_name_normalized and item_category_normalized enable cross-business comparison (e.g. compare "cheese pizza" prices across all metros)
• Freshness guarantee — only observations from the past 180 days; stale data is excluded on each export
• Quality-filtered — confidence > 0.3, zero quality flags, price range $1.00–$500 per item
• Source-typed — source_kind distinguishes website, pdf, and fax submissions

Best use cases

• Restaurant price benchmarking across metros, cuisines, and item categories
• Menu price comparison for chains, independents, and local market studies
• Food inflation and regional price dispersion analysis
• RAG and structured retrieval for restaurant/menu price assistants
• Model evaluation or fine-tuning for menu understanding, price extraction, and normalization

Quick Start

Python — datasets library

from datasets import load_dataset
import pandas as pd

ds = load_dataset("grichardomi/restaurant-prices", split="train")
df = pd.json_normalize(ds.to_list())

# Median price by metro
print(df.groupby("business.metro")["observation.price_usd"].median().sort_values())

pandas

import pandas as pd

df = pd.json_normalize(
    pd.read_json("hf://datasets/grichardomi/restaurant-prices/train.jsonl", lines=True).to_dict("records")
)

# Price spread by category across metros
print(df.groupby(["business.metro", "observation.item_category_normalized"])["observation.price_usd"]
        .agg(["median", "count"])
        .query("count >= 5")
        .sort_values("median", ascending=False))

DuckDB

INSTALL httpfs; LOAD httpfs;

SELECT
  business.metro,
  observation.item_category_normalized  AS category,
  ROUND(MEDIAN(observation.price_usd), 2) AS median_price_usd,
  COUNT(*)                               AS observations
FROM read_ndjson(
  'https://huggingface.co/datasets/grichardomi/restaurant-prices/resolve/main/train.jsonl'
)
GROUP BY 1, 2
HAVING observations >= 5
ORDER BY observations DESC
LIMIT 20;

Coverage

Metro areas	128
Price observations	10,000
Sources	website, pdf
Window	past 180 days
Price range per item	$1.00 – $500
Min confidence	0.3

Schema

Each row is a JSON object in train.jsonl (NDJSON):

{
  "schema_version": "seed-dataset-v1",
  "record_type": "seed_observation",
  "source_kind": "website | pdf | fax",
  "observation_id": 123456,
  "business": {
    "name": "Joe's Pizza",
    "url": "https://joespizza.com",
    "metro": "new-york",
    "city": "New York",
    "state": "NY",
    "category": "pizza",
    "address": "123 Main St, New York, NY 10001",
    "lat": 40.7128,
    "lon": -74.006
  },
  "snapshot": {
    "detected_at": "2026-06-15T14:22:00.000Z",
    "menu_page_url": "https://joespizza.com/menu",
    "avg_price_usd": 14.5
  },
  "observation": {
    "price_cents": 1450,
    "price_usd": 14.5,
    "currency": "USD",
    "price_raw": "$14.50",
    "size": "Large",
    "modifier": null,
    "observed_at": "2026-06-15T14:22:00.000Z",
    "confidence": 0.95,
    "extraction_method": "schema_org",
    "is_bulk_price": false,
    "quality_flags": [],
    "item_name_raw": "Cheese Pizza - Large",
    "item_name_normalized": "cheese pizza",
    "item_category_normalized": "pizza"
  }
}

Key fields

Field	Type	Description
business.metro	string	Metro slug (e.g. new-york, los-angeles)
business.lat / business.lon	float	Business coordinates
observation.price_usd	float	Price in USD
observation.item_name_normalized	string	Normalized item name for cross-business comparison
observation.item_category_normalized	string	Normalized category (e.g. pizza, burger, taco)
observation.confidence	float	Extraction confidence 0–1
observation.is_bulk_price	bool	True if price covers multiple items (e.g. party tray)
source_kind	string	website, pdf, or fax

Data Collection

Prices are extracted from publicly accessible restaurant websites, PDFs, and approved fax submissions using structured data parsing, HTML extraction, OCR, and normalization. Only records with zero quality flags and confidence > 0.3 are included. User-private URLs and internal ownership fields are excluded from this public export.

Contact & Commercial Access

Homepage: marketpulse.report Commercial licensing, bulk access, API inquiries: data@getmarketpulse.com Point of contact: MarketPulse Data Team

This repo contains a public sample (10,000 observations) for evaluation, schema exploration, and research. For the full rolling dataset, bulk downloads, or API access, contact us or request access via the full dataset repo.

License

CC BY 4.0 — free to use with attribution: MarketPulse (marketpulse.report) · View Collection

Citation

@dataset{marketpulse2026,
  title={MarketPulse Restaurant Price Dataset},
  author={MarketPulse},
  year={2026},
  url={https://huggingface.co/datasets/grichardomi/restaurant-prices},
  license={CC BY 4.0}
}