EDIT: Research originally done as part of a HSA, I’ll publish it here because why not, may be useful.

Best

• Demand Forecast for Optimized Inventory Planning | Kaggle
  • Literally DMC 2020: DMC 2020 – DATA MINING CUP
  • We have orders for ItemID, for each a salesPrice
  • items.csv with item/brand/manufacturer, customerRating and three category attributes (‘categorical/hierarchical’?).
• Retail Store Sales Transactions (Scanner Data) | Kaggle
  • One year: 2016
  • Really nice!
  • Date, CustomerID, item, category, quantity, ~price

  • The anonymized dataset includes 64.682 transactions of 5.242 SKU’s sold to 22.625 customers during one year.

• Retail Data Set | Kaggle
  • Date, Article+quantity, price, discount, customer ID
  • 29k rows, 2019-2023
• Demand Forecasting | Kaggle
  • 2011-2013, ~2k stores
  • for each week: product id, units sold, total price + base price (?), whether it as featured and displayed (?)

Interesting

• Walmart Dataset | Kaggle
  • 45 stores, weekly sales in total, not per-item!
  • Holiday week or not + Temperature / fuel price / CPI / Unemployment
  • 2010-2012, of course USA
• Store Item Demand Forecasting Challenge | Kaggle
  • Daily sales of 50 items across 10 stores, no categories
  • 2013-2017
  • time-based train/test split
  • Used here: Machine Learning for Retail Demand Forecasting | by Samir Saci | Towards Data Science
• Predicting the sales of products of a retail chain | Kaggle
  • 2012-2014, product/dept and one category
    • 333 stores in 3 Indian states
  • Has prices or each product by outlet and week
• Supermarket sales | Kaggle
  • per-category data!
    • category (“Electronics”), quantity bought, price, datetime of sale
    • Price, + tax and A LOT of similar info
      • • “Gross income” whatever this means
    • Branch, city
    • Customer gender/membership
    • Customer rating of shopping experience
  • Jan-Mar 2019

Not too interesting

• Retail Sales Dataset 2018-2022 | Kaggle
  • Monthly sales data, article and “group” (=category), avg. price etc.
  • 2018-2022, no stores/cities, “global retail company” - could be aggregated worldwide?
• Pakistan’s Largest E-Commerce Dataset | Kaggle
  • Everything we might want, but E-commerce and Pakistan
  • Categories and items not anonymized!
  • Looks messy
  • Completed/cancelled orders
• Retail Demand Forecasting Dataset | Kaggle
  • Unknown/anonymous
  • Products belonging to one of 70k (long-tailed disr.) categories
  • warehouse demand
  • Jan-Nov 2016
  • For each week: it’s a state/school holiday, promo, etc.
  • Weekly SKU level Product Sales Transactions | Kaggle
    • for each article - number of sales in 6 weeks + price + store
• Forecasts for Product Demand | Kaggle
  • More about manufacturing demand and bulk things than retail
• UCI Machine Learning Repository: Demand Forecasting for a store Data Set
  • Meal delivery company
  • Category, sub-category, price, discount, featured on homepage etc.
  • Has per-meal metadata (cuisine etc.)
  • Hierarchical sales data of an Italian grocery store - Mendeley Data
    • 2014-2018, pasta sold in an italian store, + presence/absence of promotion
  • Australia Grocery Product Dataset | Kaggle
    • No demand, just not-anonimysed prodcts with category/subcategory and price available in Australia on a per-city-postal-code basis and whether it’s in stock or not at that time
    • May be useful as taxonomy or something?..
• Superstore Dataset | Kaggle
  • FAKE Tableau sample dataset, but is 10/10 what we look for
  • Has everything, not anonymized - product + quantity, category+sub-category, price, discount
    • order/shipping date, customer name/segment/city
  • 2014-2017

TODO

• Continue looking through this: Search | Kaggle
• Corporación Favorita Grocery Sales Forecasting | Kaggle
  • Ecuador, the usual

  • dates, store and item information, whether that item was being promoted, as well as the unit sales. Additional files include supplementary information that may be useful in building your models.

  • TODO I get an error when I “accept” the terms of the competition in two browsers, can’t look inside
• Retail Sales Forecasting | Kaggle
  • “Brazilian top retailer”, many stores etc.
  • Four columns, ‘venda’ and ’estoque’ ?..
    • I guess all categories and items summed up
  • 2014-2016

Sources / useful search terms

• retail | grocery, demand | sales
• SKU is a term of art, basically ArticleID
• “merchandise planning”
• Kaggle category ‘retail and shopping’: Find Open Datasets and Machine Learning Projects | Kaggle
• SKU: Search | Kaggle
• Dataset Search
• Dataset Search

Random

• Book about forecasting! Chapter 10 Forecasting hierarchical or grouped time series | Forecasting: Principles and Practice (2nd ed)