User Guide

Complete guide to using ACRO for statistical disclosure control.

User Guide

• Getting Started
  • What is ACRO?
  • Key Concepts
  • Basic Workflow
  • Mitigation Strategies
  • How ACRO’s Ontology-Driven Architecture Works
  • Federated Mode
  • Installation Requirements
  • Next Steps
• Core Concepts
  • Principles-Based SDC
  • The New Ontology-Driven Architecture (v1.0.0+)
  • Disclosure Control Methods
  • Federated Mode
  • How Categories Are Preserved
  • The Process at a Glance
  • Best Practices
• Configuration
  • Configuration Methods
  • Safety Parameters
  • Configuration Precedence
  • Best Practices
• Architecture Reference
  • Background and Motivation
  • The Four JSON Knowledge Files
  • The Core Classes
  • Suppression via Data Redaction
  • Federated Evidence Serialisation
  • Regenerating the Knowledge Files
  • Adding a New Analysis Type

Overview

This user guide provides comprehensive documentation for using ACRO effectively in your research workflows. If you are new to ACRO, start with Getting Started. If you want to understand how the disclosure checking works, see Core Concepts and Architecture Reference.

Quick Navigation

• New users: Start with Getting Started

• Understand the checks: Read Core Concepts

• Configure thresholds: See Configuration

• Developers / TRE admins: See Architecture Reference

• Hands-on examples: Browse Examples

Key Topics

Data Analysis

• Cross-tabulations Safe table creation with automatic disclosure control

• Statistical modelling Regression analysis with privacy protection

• Summary statistics Descriptive statistics with safety checks

• Data visualisation Safe histograms and survival plots

Mitigation Strategies

• Suppression Remove records that fall into disclosive cells, then rerun the table

• Rounding Round all cell values to the nearest multiple of a configurable base

Configuration

• Safety parameters Customising disclosure thresholds via YAML

• Federated mode Evidence-only mode for use with a trusted aggregator

• Ontology knowledge base How the JSON lookup files are generated and updated

Integration

• Python workflows Jupyter notebook integration

• R integration Using the ACRO-R package

• Stata workflows Statistical software integration

Best Practices

1. Use suppress=True or enable_rounding() in production environments.

2. Review all review and fail outputs before calling finalise().

3. Name your outputs with rename_output() for easier output checker review.

4. Document your analysis add comments and exceptions as you go.

5. Test with synthetic data before running on sensitive datasets.

Getting Help

• API Reference Complete API reference

• GitHub Issues Report bugs

• Discussions Community support