FreshRetailNet-50K
收藏资源简介：
FreshRetailNet-50K是一个大规模的生鲜零售库存缺货标注的审查需求数据集，旨在解决生鲜零售中需求估计的挑战。该数据集包含来自18个主要城市898家商店的50,000条详细的每小时销售数据，涵盖了863种生鲜SKU，并仔细标注了缺货事件。独特的每小时库存状态记录，加上丰富的上下文协变量，包括促销折扣、降雨量和时间特征，使得研究能够超越现有解决方案。数据集的开放性质为需求插补、生鲜库存优化和因果零售分析等领域的创新研究提供了新的方向。
FreshRetailNet-50K is a large-scale annotated dataset designed for stockout audit in fresh retail inventory management, aiming to tackle the core challenges of demand estimation in fresh retail operations. This dataset encompasses 50,000 detailed hourly sales records collected from 898 stores across 18 major cities, covering 863 fresh produce SKUs, with meticulously annotated stockout events. The unique hourly inventory status records, combined with rich contextual covariates including promotional discounts, rainfall data and temporal features, enable researchers to develop solutions that outperform existing approaches. The open-access nature of this dataset provides new research directions for innovative studies in fields such as demand imputation, fresh inventory optimization and causal retail analysis.
FreshRetailNet-50K 数据集概述
数据集基本信息
- 语言: 英语
- 许可证: CC BY 4.0
- 任务类别: 时间序列预测
- 标签: 生鲜零售、需求截断、每小时库存状态
- 规模: 1M < n < 10M
- 配置: 默认配置包含训练集和评估集
数据集详情
- 名称: FreshRetailNet-50K
- 概述: 首个大规模生鲜零售领域需求截断估计基准数据集，包含约20%自然发生的缺货数据
- 数据量: 50,000个商店-产品90天时间序列
- 覆盖范围: 18个主要城市898家商店的863种易腐SKU
- 独特特征: 每小时库存状态记录，丰富的上下文协变量（促销折扣、降水等时间特征）
数据字段
| 字段 | 类型 | 描述 |
|---|---|---|
| city_id | int64 | 编码城市ID |
| store_id | int64 | 编码商店ID |
| management_group_id | int64 | 编码管理组ID |
| first_category_id | int64 | 编码一级类别ID |
| second_category_id | int64 | 编码二级类别ID |
| third_category_id | int64 | 编码三级类别ID |
| product_id | int64 | 编码产品ID |
| dt | string | 日期 |
| sale_amount | float64 | 全局归一化后的日销售额 |
| hours_sale | Sequence(float64) | 全局归一化后的每小时销售额 |
| stock_hour6_22_cnt | int32 | 6:00-22:00缺货小时数 |
| hours_stock_status | Sequence(int32) | 每小时缺货状态 |
| discount | float64 | 折扣率 |
| holiday_flag | int32 | 假期标志 |
| activity_flag | int32 | 活动标志 |
| precpt | float64 | 总降水量 |
| avg_temperature | float64 | 平均温度 |
| avg_humidity | float64 | 平均湿度 |
| avg_wind_level | float64 | 平均风力 |
层级结构
- 仓库: city_id > store_id
- 产品类别: management_group_id > first_category_id > second_category_id > third_category_id > product_id
使用方式
python from datasets import load_dataset dataset = load_dataset("Dingdong-Inc/FreshRetailNet-50K") print(dataset)
许可证与使用条款
- 许可证: 知识共享署名4.0国际许可协议 (CC BY 4.0)
- 数据开发者: Dingdong-Inc
- 发布日期: 2025年5月8日
- 数据版本: 1.0
预期用途
- 用于改进潜在需求恢复和需求预测技术
- 用户需自行检查数据集许可证是否适用于预期目的
引用
@article{2025freshretailnet-50k, title={FreshRetailNet-50K: A Stockout-Annotated Censored Demand Dataset for Latent Demand Recovery and Forecasting in Fresh Retail}, author={Yangyang Wang, Jiawei Gu, Li Long, Xin Li, Li Shen, Zhouyu Fu, Xiangjun Zhou, Xu Jiang}, year={2025}, eprint={2505.16319}, archivePrefix={arXiv}, primaryClass={cs.LG}, url={https://arxiv.org/abs/2505.16319}, }
相关资源
- 1FreshRetailNet-50K: A Stockout-Annotated Censored Demand Dataset for Latent Demand Recovery and Forecasting in Fresh Retail中山大学 · 2025年
