Hugging Face's logo

Datasets:
iizy
/
calcfi-open-data

Tasks:
Tabular Regression
Time Series Forecasting
Languages:
English
Size:
100K<n<1M
Tags:
finance
economics
household-finance
macro
time-series
open-data
License:
Dataset card Files
xet
 Community
calcfi-open-data / docs /datahub-batch.md
iizy's picture
iizy
Add STATUS.md + DataHub batch publishing guide
991492f
preview code
|
raw
history blame contribute delete
2.34 kB

DataHub.io batch publishing

DataHub.io requires one post per dataset, but all posts can point at the same GitHub repo (jerehere/calcfi-open-data) with different Root Dir values.

Form fields (every post uses these)

  • Post Type: Standard Dataset
  • GitHub Account: jerehere
  • Repository: calcfi-open-data
  • Branch: main
  • Root Dir: see batch list below (paste exactly, no trailing slash)

Batch list — 25 datasets with full history (publish these first) 

datasets/30-year-fixed
datasets/15-year-fixed
datasets/10-year-treasury
datasets/federal-funds-rate
datasets/prime-rate
datasets/2-year-treasury
datasets/cpi
datasets/pce
datasets/unemployment-rate
datasets/labor-force-participation
datasets/avg-hourly-earnings
datasets/crude-oil-wti
datasets/crude-oil-brent
datasets/us-regular-gas-price
datasets/usd-eur
datasets/usd-gbp
datasets/usd-jpy
datasets/credit-card-apr
datasets/personal-loan-rate-24mo
datasets/copper-price
datasets/corn-price
datasets/us-gdp-per-capita
datasets/world-unemployment
datasets/eurozone-gdp-per-capita

Defer until CalcFi history accumulates (9 datasets, currently snapshot-only) 

datasets/bitcoin
datasets/ethereum
datasets/solana
datasets/savings-national-rate
datasets/money-market-national-rate
datasets/cd-3mo-national-rate
datasets/cd-12mo-national-rate
datasets/cd-60mo-national-rate
datasets/auto-loan-rate-48mo
datasets/world-inflation

Recommended publishing cadence

  • Day 1: top 5 highest-traffic series (30-year-fixed, cpi, federal-funds-rate, unemployment-rate, 10-year-treasury)
  • Day 2-3: remaining 20 history-bearing series
  • Day 7+: re-run node scripts/refresh.mjs locally to see which deferred series now have history; publish those

After publishing

Each DataHub post auto-syncs from GitHub on every push. The daily refresh workflow will keep all datasets current with zero manual work.

Other registries to mirror

Same repo, different platforms:

  • Hugging Face Datasets: huggingface-cli upload jerehere/calcfi-30y-mortgage datasets/30-year-fixed/data.csv
  • Kaggle: create new dataset, upload data.csv, paste README content
  • Zenodo: connect repo, tag a release (gh release create v0.1.0), Zenodo creates DOI automatically
  • re3data.org: registry submission form once 3+ external mirrors exist