README.md

metadata

license: mit
task_categories:
  - time-series-forecasting
tags:
  - time-series
  - forecasting
  - ts-arena
  - long-horizon
size_categories:
  - 10K<n<100K

Traffic Dataset

Part of the TS Arena benchmarking suite for time-series foundation models.

Description

Road occupancy rates from 862 sensors in San Francisco Bay Area

Statistics

• Total samples: 17544
• Features: 862
• Train samples: 12280
• Validation samples: 1755
• Test samples: 3509
• Frequency: 1 hour

Preprocessing

• Standard normalization (zero mean, unit variance)
• Scaler fitted on training data only
• Train/Val/Test split: 70%/10%/20%

Usage

from datasets import load_dataset

# Load a specific split
train = load_dataset("ts-arena/traffic", split="train")
val = load_dataset("ts-arena/traffic", split="validation")
test = load_dataset("ts-arena/traffic", split="test")

Source

Autoformer Dataset Collection

License

MIT