FreshRetailNet-50K: Fresh Retail Demand with Stockout Annotations

Provider
Dingdong Inc.
License
CC-BY-4.0 (source)
Size
5 files
Format
Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/Dingdong-Inc/FreshRetailNet-50K)
Access
Free and public. No account with the data provider is needed.
Category
Machine Learning & AI
Source
https://arxiv.org/abs/2505.16319

About FreshRetailNet-50K: Fresh Retail Demand with Stockout Annotations

FreshRetailNet-50K, described by its authors as the first large-scale benchmark for censored demand estimation in fresh retail. It contains 50,000 store-product time series covering 90 days of detailed hourly sales from 898 stores in 18 major cities and 865 perishable SKUs, with stockout event annotations; about 20% of the data are organically occurring stockouts, so observed sales understate true demand. Rows carry encoded city, store, management group, three-level category and product ids, the date, normalised daily sale amount, hourly sales and hourly stock status, and covariates: discount rate, holiday and activity flags, precipitation, average temperature, humidity and wind level. Listing data/ (measured 2026-11-14) shows train.parquet (about 106 MB) and eval.parquet (about 8 MB). Uses include latent demand recovery under stockouts, demand forecasting for perishable goods, inventory and replenishment optimisation and time series imputation. Load with Hugging Face datasets or pandas; the GitHub repository has the baseline pipeline. Licensed CC BY 4.0 per the card; cite the FreshRetailNet-50K technical report.

How to load FreshRetailNet-50K: Fresh Retail Demand with Stockout Annotations in Python

from datasets import load_dataset
ds = load_dataset("Dingdong-Inc/FreshRetailNet-50K")
print(ds)

Data structure

File types in sampled files: Parquet

Files

Example file: data/eval.parquet

Analyze this dataset without downloading it

On Scigantic, FreshRetailNet-50K: Fresh Retail Demand with Stockout Annotations opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Machine Learning & AI datasets