README.md

metadata

license: other
task_categories:
  - time-series-forecasting
task_ids:
  - univariate-time-series-forecasting
  - multivariate-time-series-forecasting
annotations_creators:
  - no-annotation
source_datasets:
  - original
tags:
  - time-series
  - forecasting
  - benchmark
  - fev
  - tsfile
  - apache-tsfile
  - modality:timeseries
  - Time-series
  - format:tsfile
  - arxiv:2509.26468
size_categories:
  - 100K<n<1M
pretty_name: rossmann (TsFile format)
configs:
  - config_name: default
    data_files:
      - split: train
        path: '**/*.tsfile'

rossmann (TsFile format)

This repository contains time-series forecasting data stored in Apache TsFile format.

Summary

• FEV subset: rossmann
• Unified source collection: autogluon/fev_datasets
• Original source: https://www.kaggle.com/competitions/rossmann-store-sales
• Paper / citation: [21]
• Series: 1,115
• Modalities: Time-series
• TsFile rows (flattened observations): 8,242,080
• Frequencies: 1D, 1W
• TsFile files: 3
• Time precision: milliseconds (INT64).

Licensing and citation requirements follow the original source. This repository does not claim ownership of the original data.

Dataset Statistics

Frequency	Series	Median series length	TsFile rows (observations)	Dynamic columns	Static columns	Data files
1D	1,115	942	7,352,310	7	10	1D/1D_1..1D_2.tsfile (2 shards)
1W	1,115	133	889,770	6	10	1W/1W.tsfile

Files

The Hugging Face dataset card YAML points configs.data_files to all *.tsfile files in this repository.

• 1D/1D_1.tsfile
• 1D/1D_2.tsfile
• 1W/1W.tsfile

TsFile Storage Model

• Each original series (id) is stored as one TsFile device.
• Static covariate columns are stored as TAG columns: Store, StoreType, Assortment, CompetitionDistance, CompetitionOpenSinceMonth, CompetitionOpenSinceYear, Promo2, Promo2SinceWeek, Promo2SinceYear, PromoInterval.
• Time-varying targets and dynamic covariates are stored as FIELD measurements.
• Source timestamp values are mapped to the TsFile Time column as millisecond timestamps.
• Table name(s): rossmann_1D, rossmann_1W.

Column Schema

Column	Role	TsFile type
Time	Time column	INT64
id	TAG (device dimension)	STRING
Store	TAG (device dimension)	DOUBLE
StoreType	TAG (device dimension)	STRING
Assortment	TAG (device dimension)	STRING
CompetitionDistance	TAG (device dimension)	DOUBLE
CompetitionOpenSinceMonth	TAG (device dimension)	DOUBLE
CompetitionOpenSinceYear	TAG (device dimension)	DOUBLE
Promo2	TAG (device dimension)	DOUBLE
Promo2SinceWeek	TAG (device dimension)	DOUBLE
Promo2SinceYear	TAG (device dimension)	DOUBLE
PromoInterval	TAG (device dimension)	STRING
DayOfWeek	FIELD (measurement)	FLOAT
Sales	FIELD (measurement)	FLOAT
Customers	FIELD (measurement)	FLOAT
Open	FIELD (measurement)	FLOAT
Promo	FIELD (measurement)	FLOAT
StateHoliday	FIELD (measurement)	STRING
SchoolHoliday	FIELD (measurement)	FLOAT

Note: 2230 original id values contained invalid identifier characters and were normalized to valid device names, for example 1→_1, 2→_2, 3→_3.

Conversion Notes

• The source FEV format stores each time series as one nested row containing id, timestamp[], and target or covariate arrays.
• The TsFile conversion flattens those nested arrays into long rows. Therefore, the TsFile rows values above correspond to the number of timestamped observations after flattening.
• TAG columns identify the device and static metadata. FIELD columns contain values that change over time.
• Large logical tables may be split into multiple .tsfile shards such as <name>_1.tsfile, <name>_2.tsfile, and so on. Shards listed for the same frequency belong to the same logical table.

Reading Example

from tsfile import TsFileReader

reader = TsFileReader("1D/1D_1.tsfile")
schemas = reader.get_all_table_schemas()
# Table name(s): rossmann_1D, rossmann_1W