Prompt
What dataset should I use for time series forecasting?
Latest observation
Oct 16, 2026
OpenAI APIWeb search: off
It depends on your goal, but here are good choices by use case:
If you want a classic benchmark
- Air Passengers — tiny, simple, good for learning basics.
- Sunspots — seasonal-ish long-term patterns.
- M4 / M5 datasets — very popular forecasting benchmarks with many series.
If you want real-world business forecasting
- Store sales / retail demand — great for demand forecasting.
- Electricity load datasets — useful for energy forecasting.
- Weather or temperature datasets — good for continuous forecasting problems.
- Traffic datasets — common for multivariate forecasting.
If you want to test deep learning methods
Use larger datasets with many series:
- M4/M5
- Electricity
- Traffic
- Weather
- ETT (Electricity Transformer Temperature) datasets
If you’re just starting
Use:
- Air Passengers
- Sunspots
- M4 subset
They’re easy to load, understand, and benchmark.
Best general recommendation
If you want a strong, practical dataset for experimentation, I’d suggest:
- M4 for forecasting benchmarks
- M5 for retail demand forecasting
- Electricity/Traffic/Weather for multivariate time series
If you tell me:
- whether you want single-series or multivariate
- your domain (finance, retail, energy, etc.)
- and whether you want a beginner dataset or benchmark
…I can recommend one specific dataset.