Dataset

restaurant_reviews

The restaurant_reviews dataset, published on the Hugging Face Hub by user yav1327, is a tabular and text collection of restaurant listings paired with user reviews. It is distributed in Parquet format with a single train split of 13,144 rows and an unstated license.

Open on Hugging Face

Data preview

Subset

Loading preview...

Dataset card

What this dataset contains

Each row links a restaurant record to a written review, with restaurant attributes such as RestaurantsPriceRange2, RestaurantsGoodForGroups, RestaurantsTakeOut, and GoodForKids stored inside an attributes field. Review-level fields include the review text in review, a binary label, and a star rating captured as stars. The intended task appears to be sentiment or review classification, given the presence of the label column alongside the free-text review.

Structure and fields

The schema is flat with one row per review, identified by an int64 id, a string business_id, and a restaurant name. Geographic context is provided through string fields for postal_code, state, and city, plus float64 latitude and longitude coordinates. Operational fields include review_count as int64 and is_open as int64, while the full attribute dictionary is preserved both as a nested string in attributes and as flattened columns such as attributes.RestaurantsPriceRange2, attributes.RestaurantsGoodForGroups, attributes.RestaurantsTakeOut, and attributes.GoodForKids.

How analysts use it

The dataset can be used to train or benchmark review-sentiment models, with review paired against label and stars, and to profile restaurant footprints through latitude, longitude, city, and state. Because review_count, is_open, and price-range fields are present, it also supports basic unit-economics and competitive-density studies, though the unstated license and small size should limit redistribution and large-scale commercial use.

Documentation & references 1