Prompt
What dataset should I use for demand forecasting?
Latest observation
Oct 16, 2026
OpenAI APIWeb search: off
For demand forecasting, the “best” dataset depends on what you’re forecasting and what data you already have. In practice, you usually want a dataset with these core fields:
Must-have columns
- Date / timestamp — day, week, month, etc.
- Item / product ID — what is being forecasted
- Location / store / region — if demand varies by place
- Demand / sales quantity — the target variable
- Price — if pricing affects demand
- Promotions / discounts — very important if applicable
Helpful extra features
- Holiday / seasonality flags
- Stockouts / inventory levels — to distinguish lost sales from real low demand
- Weather — useful for retail, energy, transportation
- Marketing spend / campaigns
- Lead times / supply constraints
- Competitor actions — if available
- External economic indicators — for longer-term forecasting
Common public datasets to start with
If you want to practice or benchmark models, these are widely used:
- M5 Forecasting dataset — retail sales forecasting, very popular
- Rossmann Store Sales — store-level demand with promotions and holidays
- Favorita Grocery Sales — grocery demand forecasting
- Store Item Demand Forecasting — simpler benchmark dataset
- Walmart Sales data — good for hierarchical demand patterns
Which one should you use?
- For learning / benchmarking: M5 or Rossmann
- For retail demand forecasting: M5, Favorita, or Rossmann
- For a simple first model: Store Item Demand Forecasting
- For your own business: use your internal sales + promotions + inventory data
Rule of thumb
Use the dataset that matches:
- Your forecasting granularity
(daily, weekly, monthly) - Your business structure
(product/store level, region level, etc.) - The drivers of demand
(price, promotions, holidays, weather)
If you want, I can recommend a specific dataset based on your domain, like retail, manufacturing, energy, or e-commerce.