Datasets:
item_id string | start timestamp[s] | freq string | target list | past_feat_dynamic_real list |
|---|---|---|---|---|
0 | 2018-01-01T00:00:00 | H | [679.5,3291.0,3422.25,2046.0,982.5,779.0,943.0,974.25,842.25,486.5,734.25,984.0,1224.0,2330.0,2976.0(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
1 | 2018-01-01T00:00:00 | H | [532.5,3075.0,3159.0,1792.5,996.25,768.0,924.0,1076.25,694.5,564.75,540.75,888.0,1413.0,2786.0,3505.(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
2 | 2018-01-01T00:00:00 | H | [104.0,145.0,110.5,47.25,43.25,24.5,44.33333206176758,43.0,50.66666793823242,52.0,58.0,71.6666641235(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
3 | 2018-01-01T00:00:00 | H | [90.0,249.75,227.25,135.0,89.0,69.5,72.0,75.75,78.5,70.25,91.25,107.75,123.25,228.75,321.0,207.75,10(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
4 | 2018-01-01T00:00:00 | H | [69.0,205.25,131.25,92.0,70.75,63.5,58.5,46.0,43.0,30.75,45.5,44.75,66.5,132.75,197.0,99.0,113.5,114(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
5 | 2018-01-01T00:00:00 | H | [79.0,231.5,205.5,129.25,91.25,63.0,67.25,78.0,72.0,51.75,77.25,90.25,134.5,210.25,231.0,145.25,130.(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
6 | 2018-01-01T00:00:00 | H | [155.5,504.0,352.5,203.0,134.0,107.0,144.75,153.0,129.0,100.0,99.5,154.25,237.0,306.25,466.0,310.25,(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
7 | 2018-01-01T00:00:00 | H | [311.0,797.25,727.0,386.5,273.0,236.5,245.0,238.25,169.0,173.0,190.5,253.25,350.0,533.25,677.0,431.0(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
8 | 2018-01-01T00:00:00 | H | [98.0,252.75,155.25,110.75,55.0,49.25,51.5,57.5,39.0,37.25,53.5,56.25,100.25,163.5,230.25,90.25,71.5(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
9 | 2018-01-01T00:00:00 | H | [74.0,226.0,217.5,105.25,73.0,61.5,60.0,65.25,44.75,59.5,57.5,68.25,91.75,162.75,206.5,115.5,105.5,1(...TRUNCATED) | [[178.03611755371094,707.1416625976562,688.1166381835938,388.2916564941406,236.77499389648438,177.37(...TRUNCATED) |
World2TS Data
World2TS links real-world event context to individual time series. This release
contains 31 event-annotation configurations with 4,179 series descriptions,
1,159 dataset-local events, and 56,846 event–series links. Numeric data.arrow
files are included for 26 configurations (4,169 series) whose source terms
support redistribution. The other five configurations retain their event layer
and source attribution; their numeric observations are not hosted here.
Each configuration has dataset_info.json, unified_events.jsonl, and
unified_event_series.jsonl. The 26 numeric configurations also have
data.arrow. Download the three JSON/JSONL files directly for an annotation-only
configuration. The Hub load_dataset configurations above cover the 26 Arrow files.
The event layer is retrospective context. An event link does not establish causality or guarantee that its information was available at a forecast origin. Users must check information availability before using events in prospective forecasting experiments. Event IDs are unique within a configuration, not across the whole repository.
Contents
| Configuration | Numeric source | Series | Events | Links | Arrow hosted |
|---|---|---|---|---|---|
M_DENSE |
Salesforce/lotsa_data | 30 | 53 | 1580 | yes |
PEMS_BAY |
Salesforce/lotsa_data | 325 | 22 | 1313 | yes |
SHMETRO |
Salesforce/lotsa_data | 576 | 6 | 2988 | yes |
SZ_TAXI |
Salesforce/lotsa_data | 156 | 3 | 468 | yes |
australian_electricity_demand |
Salesforce/lotsa_data | 5 | 180 | 507 | yes |
bdg-2_bear |
Salesforce/lotsa_data | 91 | 59 | 4060 | yes |
bdg-2_fox |
Salesforce/lotsa_data | 135 | 37 | 2833 | yes |
bdg-2_panther |
Salesforce/lotsa_data | 105 | 27 | 1990 | yes |
bdg-2_rat |
Salesforce/lotsa_data | 280 | 36 | 5892 | yes |
beijing_air_quality |
Salesforce/lotsa_data | 132 | 54 | 4528 | yes |
bitcoin_with_missing |
Salesforce/lotsa_data | 18 | 62 | 920 | yes |
bull |
Salesforce/lotsa_data | 41 | 41 | 1243 | yes |
cdc_fluview_ilinet |
Salesforce/lotsa_data | 316 | 19 | 2834 | yes |
covid_mobility |
Salesforce/lotsa_data | 362 | 84 | 332 | yes |
kdd_cup_2018_with_missing |
Salesforce/lotsa_data | 270 | 30 | 3798 | yes |
lcl |
Salesforce/lotsa_data | 713 | 33 | 13547 | yes |
loop_seattle |
Salesforce/lotsa_data | 323 | 29 | 4182 | yes |
los_loop |
Salesforce/lotsa_data | 207 | 12 | 484 | yes |
oikolab_weather |
Salesforce/lotsa_data | 8 | 16 | 128 | yes |
pedestrian_counts |
Salesforce/lotsa_data | 66 | 129 | 2864 | yes |
sceaux |
Salesforce/lotsa_data | 1 | 28 | 28 | yes |
smart |
Salesforce/lotsa_data | 5 | 30 | 150 | yes |
spain |
Salesforce/lotsa_data | 1 | 26 | 26 | yes |
crypto_bitcoin_daily |
coingecko | 1 | 23 | 23 | source only |
energy_rwtc_daily |
eia | 1 | 19 | 19 | yes |
finance_us_tech_daily |
yahoo | 3 | 23 | 30 | source only |
health_covid_usa_daily |
owid | 1 | 18 | 18 | source only |
macro_fred_dgs10_daily |
fred | 1 | 11 | 11 | yes |
polymarket_fed_daily |
polymarket | 1 | 24 | 24 | source only |
transport_us_proxy_daily |
yahoo | 4 | 15 | 16 | source only |
weather_isd_jfk_tmp_hourly |
noaa_isd | 1 | 10 | 10 | yes |
Licenses and source attribution
World2TS-authored dataset descriptions, series descriptions, event summaries,
and event–series links are licensed under CC BY 4.0.
Credit World2TS and cite the paper when reusing them. Source observations retain
their own terms; CC BY 4.0 does not apply to data.arrow.
- The 23 LOTSA-derived Arrow files come from
Salesforce/lotsa_data
revision
8191fd29eb5cf906ec55effca44d8059888b615d, whose dataset card specifies Apache 2.0. Cite Woo et al., Unified Training of Universal Time Series Forecasting Transformers (2024), and check any original source terms relevant to your use. energy_rwtc_dailyuses EIA data, which EIA permits to be redistributed with acknowledgment.macro_fred_dgs10_dailyuses FRED DGS10, marked public domain with citation requested; credit the Board of Governors and FRED.weather_isd_jfk_tmp_hourlyuses NOAA NCEI Global Hourly observations for the JFK station.- Raw observations from Yahoo Finance, CoinGecko, Our World in Data COVID,
and Polymarket are omitted because redistribution rights for these specific
extracts have not been established. See each configuration's
source_urlinrelease_manifest.jsonto obtain data under the source's terms.
This mixed-rights release uses the Hub metadata value license: other instead
of assigning one license to every file. release_manifest.json lists the
per-configuration rights decision. checksums.sha256 fixes every published file.
Citation
@misc{world2ts2026,
title = {World2TS: Real-World Event Context for Time-Series Forecasting},
author = {Liu, Zhiding and Yang, Jiqian and Cheng, Mingyue and Li, Zhi and Chen, Enhong},
year = {2026},
note = {Dataset and project page},
url = {https://world2ts.github.io/}
}
Project page: https://world2ts.github.io/
Code: https://github.com/yangjq713/World2TS
- Downloads last month
- -