Skip to contents

What ACRO-R Supports

Source: supports.md

This page provides a comprehensive overview of ACRO’s capabilities for researchers, developers, and non-technical stakeholders. ACRO supports a wide range of statistical analysis functions with automated disclosure control.

Supported Data Analysis Functions

Table Creation & Cross-tabulation

For Researchers: Create frequency tables and cross-tabulations with automatic cell suppression for small counts.

What ACRO Supports:

  • crosstab() - Cross-tabulation of two or more variables with frequency counting
  • pivot_table() - Spreadsheet-style pivot tables with aggregation functions
  • table() - Simple frequency tables for categorical data (R interface only)

Technical Details: - ACRO suppresses, and reports the reason why, the value of an aggregation statistic (mean, median, variance, etc.) for any cell is deemed to be sensitive - The current version of ACRO supports the three most common tests for sensitivity: ensuring the number of contributors is above a frequency threshold, and testing for dominance via N-K rules - N-K Rule : A dominance test where if the top N contributors account for more than K% of the total, the cell is considered disclosive - Frequency Threshold : Cells with fewer than a specified number of contributors are suppressed - All thresholds are configurable via YAML configuration files - For detailed methodology, see our research paper - Automatic flagging of negative or missing values for human review

Example Use Cases: - Survey response analysis by demographics - Clinical trial outcome tables - Market research cross-tabulations - Educational assessment reporting

Statistical Modeling

For Researchers: Run regression analyses with automated checks on model outputs and residual degrees of freedom.

What ACRO Supports:

  • ols() - Ordinary Least Squares linear regression
  • logit() - Logistic regression for binary outcomes
  • probit() - Probit regression for binary outcomes

Technical Details: - For regressions such as linear, probit, and logit, the tests verify that the number of residual degrees of freedom exceeds a threshold - Within the ACRO Python package, the functionality of the ‘ACRO’ class is split into a number of separate classes for maintainability and extensibility. A ‘Tables’ class contains the code necessary to perform disclosure checks on tabular data, such as crosstab. A separate ‘Regression’ class contains the code for checking regressions such as logit and probit

Example Use Cases: - Economic modeling and policy analysis - Medical research and clinical studies - Social science research - Business analytics and forecasting

Programming Language Support

Python (Primary)

For Developers: The ACRO package is a lightweight Python tool that sits over well-known analysis tools that produce outputs such as tables, plots, and statistical models

Supported Libraries: - Pandas - For data manipulation and table creation - Statsmodels - For statistical modeling and regression analysis - NumPy - For numerical computations

Python Version Support: - Python 3.10, 3.11, 3.12, 3.13

R Language

For R Users: Additional programming languages such as this R package are supported by providing front-end packages that interface with the core ACRO Python back-end

Integration Features: - Native R syntax and workflows - R Markdown and Shiny application support - Tidyverse compatibility - CRAN package availability

Getting Started in R: <span class="pre">r<span> <span class="pre">library(“acro”)<span> <span class="pre">acro_init(suppress<span> <span class="pre">=<span> <span class="pre">TRUE)<span> <span class="pre"></span>

Stata Integration

For Stata Users: ACRO is designed to let you use familiar commands in R, Stata and Python

Features: - Drop-in replacement for standard Stata commands - Familiar Stata syntax and workflows - Integration with existing Stata scripts

Disclosure Control Features

Automated Sensitivity Testing

What ACRO Checks:

For Tables: - Minimum cell counts (frequency thresholds) - Dominance rules (N-K rules for concentration) - Presence of negative or missing values

For Statistical Models: - Residual degrees of freedom thresholds - Model fit diagnostics - Parameter significance testing

For Non-Technical Users: ACRO automatically identifies when research outputs might reveal sensitive information about individuals or organizations, applying industry-standard privacy protection rules without requiring manual review of every result.

Output Management

What ACRO Provides:

  • Suppression Masks - Clear indication of which results are hidden and why
  • Summary Reports - Detailed explanation of all disclosure checks performed
  • Audit Trails - Complete record of all analysis steps and decisions
  • Exception Handling - Process for requesting release of flagged outputs

Workflow Integration: The finalise function will: Check that each output with “fail” or “review” status has an exception, if not you will be asked to enter one. Write the outputs to a directory. This directory contains everything that the output checkers need to make a decision

Supported Environments

Research Environments

Where ACRO Works: - Trusted Research Environments (TREs) - Data safe havens - Secure data centers - Academic research computing facilities - Government statistical offices - Healthcare research environments

Installation: See Installation for complete installation instructions and system requirements.

Integration Capabilities

Analysis Workflows

For Research Teams: ACRO integrates seamlessly into existing data analysis workflows, requiring minimal changes to current practices while adding comprehensive privacy protection.

Supported Workflows: - Jupyter notebook analysis - R Markdown documents - Stata do-files and scripts - Batch processing and automation - Interactive analysis sessions

Data Sources: - CSV and Excel files - Database connections - Survey data platforms - Administrative datasets - Clinical trial databases

Output Formats

What ACRO Produces: - Standard CSV files for tables - JSON metadata files for automation - Excel workbooks for human reviewers

Review Process Support: Compatible with SACRO-Viewer for interactive output review by data controllers and compliance officers.

Technical Architecture

For System Administrators:

Core Technology: Lightweight translation scripts intercept your commands and pass them through to a python ‘engine’, based on industry-standard packages that run your commands and perform statistical disclosure checks on them

System Requirements: - Python 3.10+ runtime environment - Standard scientific computing libraries (pandas, numpy, statsmodels) - Minimal computational overhead - No external network dependencies during analysis

Security Features: - Local processing only (no cloud dependencies) - Audit logging and tracking - Configurable disclosure thresholds - Role-based access controls (through integration with TRE systems)

Documentation and Support: Standard Python coding and naming practices have been used throughout. GitHub continuous integration (CI) runners automatically generate and publish API documentation using the Python docstrings written in numpydoc format

What ACRO Does NOT Support

Current Limitations: - Complex visualizations and plots (coming in future versions) - Time series analysis (specialized disclosure rules needed) - Machine learning models (use SACRO-ML for AI/ML workflows) - Real-time data streams - Distributed computing frameworks

Alternative Solutions: - SACRO-ML - For machine learning and AI model disclosure control - SACRO-Viewer - For interactive output review and approval - Traditional SDC tools - For specialized use cases not covered by ACRO