Configuration#

Comprehensive guide to configuring ACRO for your research environment and organizational policies.

Configuration Methods#

YAML Configuration Files#

Primary method for persistent configuration:

# acro_config.yaml
safe_threshold: 10
safe_dof_threshold: 10
safe_nk_n: 2
safe_nk_k: 0.85
safe_p_percent: 0.1
check_missing_values: true
survival_safe_threshold: 10

Runtime Parameters#

You can specify the configuration file and whether to suppress unsafe outputs at runtime:

import acro

# Initialize with custom parameters
session = acro.ACRO(
    config="custom_config.yaml",  # Path to your config YAML file
    suppress=True                 # Suppress unsafe outputs
)

Safety Parameters#

TRE Risk Appetite Settings#

These settings are typically controlled by TRE administrators and cannot be changed by researchers.

safe_threshold (default: 10)

Minimum number of observations required in table cells

safe_dof_threshold (default: 10)

Minimum degrees of freedom for statistical models

safe_nk_n (default: 2)

Number of largest contributors for NK dominance rule

safe_nk_k (default: 0.85)

Percentage threshold for NK dominance rule

safe_p_percent (default: 0.1)

P-percent rule threshold for concentration

check_missing_values (default: true)

Include missing values in safety calculations

survival_safe_threshold (default: 10)

Minimum observations for survival analysis outputs

zeros_are_disclosive (default: true)

Whether zero values are considered disclosive. TREs can set this to false for datasets where class disclosure is not relevant

Behavioral Settings#

These settings can be controlled by researchers to choose how they want to mitigate risk.

suppress (default: false)

Whether to suppress unsafe outputs automatically.

Configuration Precedence#

Settings are applied in order of precedence:

  1. Runtime method parameters (where supported)

  2. Configuration file settings

  3. Default values

Best Practices#

Version Control#

  • Store configuration files in version control

  • Use meaningful commit messages for config changes

Documentation#

  • Document all custom settings and their rationale

  • Maintain configuration change logs

  • Include configuration in analysis documentation

Testing#

  • Test configuration changes with sample data

  • Validate against organizational policies