Dimensions

A dimension is a raw, non-aggregated field — a category, date, or other attribute — used to group and filter metrics. Where a metric answers "how much?", a dimension answers "broken down by what?".

Defining a dimensionCopy Anchor Link

Dimensions are usually defined on a data model, close to the column they come from:

Model orders {
  // type, data source and table

  dimension created_at {
    label: 'Created At'
    type: 'datetime'
    definition: @sql {{ #SOURCE.created_at }};;
  }

  dimension status {
    label: 'Status'
    type: 'text'
    definition: @sql {{ #SOURCE.status }};;
  }
}

A dataset can also add dimensions that combine fields across its included models. Set model to the model the dimension belongs to, and write the definition in AQL so it can reach the other models' fields:

Dataset sales {
  // models and relationships

  dimension line_value {
    model: order_items
    label: 'Line Value'
    type: 'number'
    definition: @aql order_items.quantity * products.price;;
  }
}

Queries use it like any field of that model: order_items.line_value.

Common typesCopy Anchor Link

• date and datetime — supports querying at different grains (day, week, month, …) without redefining the dimension, as in orders.created_at | month().
• text — a discrete attribute like status or plan, used for filtering and grouping.
• number — numeric values such as IDs, quantities or prices.
• truefalse — boolean flags.

Putting it togetherCopy Anchor Link

Once a dataset has metrics and dimensions defined, anyone can query revenue grouped by orders.created_at at a month grain, filtered to one orders.status — without writing SQL or repeating the underlying joins:

anfra query -d sales -a 'explore {
  dimensions { month: orders.created_at | month() }
  measures { revenue: revenue }
  filters { orders.status == "delivered" }
}'