How to Create Dataset Dimensions

This guide walks you through creating dimensions for your dataset.

Prerequisites

You should have already created a dataset config. See the dataset registration tutorial for details.

Steps

1. Identify Required Dimension Records

Identify the dimension records that the project is expecting for your dataset:

  • Inspect the project’s base dimensions and records

  • Inspect the required_dimensions section of the project config for your dataset. It may specify a subset of the base dimensions

  • Resolve any discrepancies with the project coordinator

2. Record Trivial Dimensions

Record trivial dimensions in the dataset config. See dataset concepts for more information on trivial dimensions.

3. Identify Unique Records

Identify the unique records of each dimension type in your dataset. Record them in dimension record files.

4. Reference Existing Dimensions

For any dimensions that match the project, record the existing dimension ID in the dataset config.

5. Create Dimension Mappings

For any dimensions that differ from the project, create mappings. See the dataset mapping guide for details.

6. Define Dimensions in Config

Define your dimensions in the dataset config. See the dimension model reference for the complete schema.

Next Steps