Datasets (Root Data)

A Dataset is the raw, immutable source data. It is the original table of rows and columns -- imported from a CSV, API, or created manually. A dataset contains all your data as-is, without any filtering, coloring, or mapping applied. Think of it as the single source of truth.

Key characteristics:

• Contains the full, unmodified data (all rows, all columns)
• Has a detected schema (column names, types, sample values)
• Can be public (curated, available to all users) or private (workspace-specific)
• Can be reused across many selections and scenes
• Editing a dataset does not affect scenes directly -- scenes consume selections, not datasets

Selections (Prepared Data)

A Selection is a prepared view of a dataset, configured for use in a specific visualization. It takes the root dataset and applies slicing, filtering, column mapping, row selection, coloring, sorting, and geo mapping to produce the exact data a scene needs.

Key characteristics:

• Always references a parent dataset (the root data source)
• Defines which rows are included (row selection, slicing, regional filtering)
• Defines which columns map to which visualization roles (category, value, group, label, geo ID)
• Defines how rows are colored (categorical scheme, color range, or manual per-row colors)
• Defines sorting and grouping order
• Defines geo mapping (country column, admin level, focus region) for geographic scenes
• Multiple selections can be created from the same dataset, each with different configurations
• A scene can link to one or more selections (multi-dataset support)

Example Workflow

1. Import a dataset containing GDP data for 200 countries across 10 years
2. Create Selection A: Slice by year = 2024, select top 20 countries, map "Country" to category and "GDP" to value, apply a blue color gradient
3. Create Selection B: Slice by year = 2020, same 20 countries, red color gradient
4. Link both selections to a scene for a side-by-side comparison

The Datasets page lists all available datasets in your workspace and provides access to public datasets.

The Dataset Detail page shows comprehensive information about a single dataset, whether public or private.

The Selection Editor is where you configure how a dataset's root data is prepared for a visualization scene. Every setting here creates the bridge between raw data and the 3D scene.

Selection Name

The selection name is editable via an inline edit button. Click Save in the breadcrumb header to persist all changes.