Hubverse Forecast Prep

What each field means

task ID columns

The hub-specific columns that identify the forecast task, such as reference_date, target, horizon, target_end_date, and location.

location

The hub's identifier for the geographic unit, such as a FIPS code or a custom location ID. Values are kept as text so leading zeros are preserved.

output_type

The forecast representation, such as mean, median, quantile, pmf, or sample.

output_type_id

The representation-specific identifier, such as a quantile probability, sample index, or probability-mass category.

value

The submitted prediction for the specified task and output representation.

How preparation works

Add dataUpload Excel, CSV, TSV, JSON, text, paste, or manually enter rows.
Choose a hubApply a bundled profile or load the hub's current tasks.json, then match fields.
Assign tasksSet targets and crosswalk source locations to hub IDs.
Review checksResolve invalid dates, missing values, and duplicates.
DownloadSave a formatted forecast CSV and optional Python or R code.

Use the hub's exact setup

The selector includes the 30 public hubs in the Hubverse directory, grouped as active, training, model development, and archival entries (directory updated 31 August 2026). Private directory entries are deliberately omitted.

If a hub provides hub-config/tasks.json, load it during mapping when you need rules newer than the bundled profile. The tool reads task-ID columns, target and location values, output types, output-type IDs, and value rules, then adjusts the form and checks.

Only forecast fields required for the selected hub are exported. The standard output_type, output_type_id, and value columns are always retained, and recognized task dates are written as ISO YYYY-MM-DD.

A hub can still define additional rules that only its official validation workflow knows. Treat this browser check as preparation—not a replacement for final hub validation.

Private by design

All parsing, cleaning, checking, and file creation happen inside this browser tab. This offline HTML makes no automatic network requests and has no analytics or external code dependencies.

Read Hubverse model-output guidance

Turn forecast files into hub-ready model output.

Add your forecast file

Choose the hub and match your columns

Choose the destination hub

Match the forecast columns

Reusable pipeline

Confirm the cleaning defaults

Assign forecast tasks and hub locations

Assign one target to every row

Match source locations to hub IDs

Review the checks

Check details

Output preview

Your Hubverse file is ready

A focused bridge from source forecasts to Hubverse.

What each field means

How preparation works

Use the hub's exact setup

Private by design

Bring in almost any table