---
license: cc-by-4.0
task_categories:
- time-series-forecasting
tags:
- tsfile
- timeseries
- time-series
- forecasting
pretty_name: Traffic (TsFile)
size_categories:
- 100K<n<1M
configs:
- config_name: default
  data_files:
  - split: train
    path: traffic.tsfile
---

# Traffic (TsFile)

Apache TsFile version of the `traffic` long-term-forecasting subset of
[`thuml/Time-Series-Library`](https://huggingface.co/datasets/thuml/Time-Series-Library).

## Overview

Hourly road-occupancy rates from 862 sensors on San Francisco Bay Area freeways.

- **Rows:** 17,544 time steps.
- **Channels:** 862 numeric variables.
- A single multivariate series; no TAG/device dimension.

## Schema (TsFile structure)

- **Time** (INT64, milliseconds) — parsed from the source `date` column.
- **FIELD** (862 columns, FLOAT) — the measured channels. Channel names are
  made schema-safe (e.g. `%`→`pct`, spaces→`_`, leading digits prefixed `c_`).

The source `date` string is dropped (losslessly encoded into Time). No rows or
numeric columns are dropped.

## Usage

Read the `.tsfile` files with the Apache TsFile Java or Python SDK.

## Source & license

- Original dataset: https://huggingface.co/datasets/thuml/Time-Series-Library (subset `traffic`)
- Author / publisher: thuml (Tsinghua University)
- Paper: https://arxiv.org/abs/2407.13278
- License: CC BY 4.0