Skip to content

Subject Areas & Datasets

Last updated: 2026-04-27 Tags: subject area, dataset, XSA, measures, attributes, dimensions, data blending

📖 Full Oracle Documentation: Visualizing Data and Building Reports · Building Semantic Models

Summary

OAC exposes data to users through two mechanisms: Subject Areas (enterprise-grade, defined in the Semantic Model) and Datasets (self-service, uploaded or queried directly). Subject Areas are the backbone of Classic Analyses and Dashboards; Datasets power Workbooks and can be joined together through data blending.

Subject Areas

What Is a Subject Area?

A Subject Area is a curated, business-friendly view of data defined in the Presentation Layer of the Semantic Model. It groups related dimensions and measures into folders, presenting a consistent, governed view to all users.

Subject Area: "Sales Performance"
  ├── Folder: Time
  │     ├── Year
  │     ├── Quarter
  │     └── Month
  ├── Folder: Product
  │     ├── Category
  │     └── Product Name
  ├── Folder: Customer
  │     └── Region
  └── Folder: Facts - Sales
        ├── Revenue       [MEASURE]
        ├── Units Sold    [MEASURE]
        └── Avg Price     [MEASURE]

Column Types

Type Description Example
Attribute Descriptive / dimensional Customer Name, Region, Product
Measure Numeric, aggregatable Revenue, Units, Count
Hierarchy Level Part of a drill path Year > Quarter > Month
  • Classic Analyses: Subject Area panel → drag columns to Criteria tab
  • Workbooks: Add Data → Subject Area → choose columns
  • Logical SQL: Reference as "Subject Area"."Folder"."Column"

Datasets

What Is a Dataset?

A Dataset is a self-service data source created by a user. It can be:

  • An uploaded file (Excel, CSV)
  • A query against a database connection
  • A result from a Data Flow
  • A blend of multiple sources

Dataset vs. Subject Area

Aspect Dataset Subject Area
Created by Any user Data modeler / Admin
Governed No (self-service) Yes (enterprise)
Joins User-defined (data blending) Pre-defined in Semantic Model
Used in Workbooks Workbooks, Analyses, Dashboards
Data location OAC in-memory cache Live query to DB via BI Server
Row limit Configurable (default 250k) No limit (live query)

Creating a Dataset

  1. Home → Create → Dataset
  2. Choose source: Connection, File, Subject Area, etc.
  3. Select tables/sheets
  4. Optionally transform in the Data Preparation editor
  5. Save → Dataset available in Workbooks

Dataset Columns

  • All columns start as attributes
  • User can change to Measure with aggregation rule
  • User can set data type, format, and aggregation
  • Hidden columns possible

Extended Subject Areas (XSA)

XSAs expose Datasets as queryable sources via Logical SQL using the XSA() function.

-- Query a dataset via Logical SQL
SELECT XSA('admin'.'sales_data')."Orders"."Customer",
       SUM(OVERRIDEAGGR(XSA('admin'.'sales_data')."Orders"."Revenue"))
FROM XSA('admin'.'sales_data')
FETCH FIRST 100 ROWS ONLY

XSA Naming

  • XSA('namespace'.'dataset_name')
  • Namespace = the username who owns the dataset (or a shared namespace)
  • Dataset name = the dataset's technical name (not display name)

💡 Tip: Use discover_data (OAC MCP tool) to discover XSA names and subject areas available in your instance.

Data Blending

Blend two or more Datasets (or Datasets + Subject Areas) in a single Workbook.

How Blending Works

  • Each data source gets its own "layer"
  • OAC joins them client-side on matching column names or user-defined join keys
  • Result: a merged view for visualization

Setting Up Blends

  1. In Workbook → Data panel → Add second dataset
  2. OAC auto-detects matching columns (same name, same type)
  3. Adjust match columns manually if needed
  4. Choose join type: Inner, Left Outer, Right Outer, Full Outer

⚠️ Warning: Data blending joins happen in memory in the browser. Very large datasets can cause performance issues. Consider using Data Flows to pre-join data server-side.

Data Preparation (Dataset Editor)

When creating or editing a dataset, the Data Preparation editor allows:

  • Rename columns
  • Change data type (String, Number, Date)
  • Hide columns
  • Add enrichments: geographic encoding, date extraction (year, quarter, month)
  • Add calculations: formula-based derived columns
  • Recommendations: OAC suggests data quality fixes (nulls, format issues)