Reporting

SecureML provides powerful reporting tools to document privacy measures, generate compliance documentation, and create audit trail reports. These tools help organizations demonstrate compliance with privacy regulations and maintain transparency in their AI systems.

Core Concepts

Report Generation: Creating structured documentation of privacy measures, compliance checks, and audit trails.

Report Formats: Output reports in different formats including HTML and PDF.

Visual Elements: Including charts, graphs, and other visual elements to aid in understanding complex privacy data.

Compliance Documentation: Generating reports specifically designed to demonstrate regulatory compliance.

Basic Usage

Generating Compliance Reports

Generate comprehensive reports from compliance checks:

from secureml import check_compliance
import pandas as pd

# Load your data
data = pd.read_csv("sensitive_data.csv")

# Check compliance
report = check_compliance(data, regulation="GDPR")

# Generate an HTML report
report.generate_report("gdpr_compliance.html")

# Generate a PDF report (requires WeasyPrint)
report.generate_report("gdpr_compliance.pdf", format="pdf")

Compliance reports include:

Overall compliance status
Detailed issues and warnings
Recommendations for addressing compliance gaps
Visual breakdown of compliance issues by severity

Generating Audit Trail Reports

Create reports from audit trails to document operations on sensitive data:

from secureml import get_audit_logs, ReportGenerator

# Retrieve audit logs for a specific operation
logs = get_audit_logs(
    operation_name="data_anonymization",
    start_time="2023-01-01T00:00:00",
    end_time="2023-01-31T23:59:59"
)

# Create a report generator
generator = ReportGenerator()

# Generate an HTML report
generator.generate_audit_report(
    logs=logs,
    output_file="audit_report.html",
    title="Data Anonymization Audit"
)

# Generate a PDF report
generator.generate_audit_report(
    logs=logs,
    output_file="audit_report.pdf",
    title="Data Anonymization Audit"
)

Advanced Techniques

Customizing Report Templates

Create custom report templates to match organizational branding:

from secureml import ReportGenerator

# Create a report generator with custom templates
generator = ReportGenerator(
    templates_dir="my_custom_templates/",
    custom_css="my_custom_style.css"
)

# Generate a report using custom templates
generator.generate_compliance_report(
    report=compliance_report,
    output_file="custom_compliance_report.html",
    logo_path="company_logo.png"
)

Adding Charts and Visualizations

Enhance reports with visual elements:

from secureml import ReportGenerator
import matplotlib.pyplot as plt
import base64
from io import BytesIO

# Create a custom chart
plt.figure(figsize=(8, 5))
plt.bar(['High', 'Medium', 'Low'], [5, 8, 3])
plt.title('Privacy Risks by Severity')

# Convert to base64
buffer = BytesIO()
plt.savefig(buffer, format='png')
plt.close()
buffer.seek(0)
chart_data = base64.b64encode(buffer.read()).decode('utf-8')

# Create a report generator
generator = ReportGenerator()

# Generate a report with the custom chart
generator.generate_compliance_report(
    report=compliance_report,
    output_file="report_with_charts.html",
    additional_context={"custom_chart": chart_data}
)

Comprehensive ML Pipeline Reporting

Generate reports for an entire ML pipeline:

from secureml import ComplianceAuditor
import pandas as pd

# Create a compliance auditor for HIPAA
auditor = ComplianceAuditor(regulation="HIPAA")

# Load dataset
data = pd.read_csv("patient_data.csv")

# Define model configuration
model_config = {
    "model_type": "RandomForest",
    "supports_forget_request": True,
    "access_controls": True,
    "parameters": {
        "n_estimators": 100,
        "max_depth": 10
    }
}

# Define preprocessing steps
preprocessing_steps = [
    {
        "name": "remove_identifiers",
        "type": "anonymization",
        "input": "raw_data",
        "output": "deidentified_data",
        "parameters": {"columns_to_remove": ["name", "ssn", "address"]}
    },
    {
        "name": "feature_selection",
        "type": "data_minimization",
        "input": "deidentified_data",
        "output": "minimal_data",
        "parameters": {"selected_features": ["age", "lab_results", "diagnosis"]}
    }
]

# Audit the entire pipeline
audit_result = auditor.audit_pipeline(
    dataset=data,
    dataset_name="patient_records",
    model=model_config,
    model_name="diagnosis_predictor",
    preprocessing_steps=preprocessing_steps,
    metadata={"data_storage_location": "US-East", "data_encrypted": True}
)

# Generate a comprehensive PDF report
auditor.generate_pdf(
    audit_result=audit_result,
    output_file="compliance_report.pdf",
    title="HIPAA Compliance Audit"
)

Integrating with Audit Trails

Use the reporting tools with audit trails to create comprehensive documentation:

from secureml import AuditTrail, ReportGenerator

# Create an audit trail for an operation
audit = AuditTrail(
    operation_name="data_anonymization",
    regulations=["GDPR"]
)

# Log dataset access
audit.log_data_access(
    dataset_name="patient_records",
    columns_accessed=["age", "gender", "zipcode", "disease"],
    num_records=5000,
    purpose="Anonymization for research"
)

# Log data transformation
audit.log_data_transformation(
    transformation_type="k_anonymity",
    input_data="Raw patient data",
    output_data="Anonymized patient data",
    parameters={"k": 5, "quasi_identifiers": ["age", "gender", "zipcode"]}
)

# Close the audit trail when done
audit.close(status="completed")

# Get the audit logs
logs = get_audit_logs(operation_name="data_anonymization")

# Generate a report from the audit logs
generator = ReportGenerator()
generator.generate_audit_report(
    logs=logs,
    output_file="anonymization_audit.pdf",
    title="Data Anonymization Audit Report"
)

Report Formats

HTML Reports

HTML reports are interactive, easy to view in browsers, and can include rich formatting:

# Generate an HTML report
report.generate_report("compliance_report.html", format="html")

PDF Reports

PDF reports are ideal for formal documentation and sharing with stakeholders:

# Install WeasyPrint for PDF support
# pip install secureml[pdf]
# On Windows, you'll also need to install GTK libraries. See:
# https://doc.courtbouillon.org/weasyprint/stable/first_steps.html#windows

# Generate a PDF report
report.generate_report("compliance_report.pdf", format="pdf")

Report Components

Compliance Reports

Compliance reports typically include:

Summary Section: Overall compliance status
Issues Section: Detailed list of compliance issues with severity ratings
Warnings Section: Potential compliance risks that need attention
Passed Checks Section: Successfully passed compliance requirements
Visual Elements: Charts showing issues by severity
Recommendations: Actionable steps to address compliance gaps

Audit Reports

Audit reports typically include:

Timeline of Events: Chronological record of operations
Operation Details: Information about each logged event
Visual Breakdown: Charts showing events by type
Metadata Section: Information about the audited operation
Regulatory Context: Applicable regulations and requirements

Integration with Command Line Interface

Generate reports using the SecureML CLI:

# Generate a compliance report from a dataset
secureml compliance check data.csv --regulation GDPR --output report.html --format html

# Generate a PDF compliance report
secureml compliance check data.csv --regulation HIPAA --output report.pdf --format pdf

# Generate an audit report from logs
secureml audit report --operation-name "data_anonymization" --output audit_report.html

Best Practices

Regular Reporting: Generate compliance and audit reports regularly, not just during audits
Version Control: Maintain report history to track compliance improvements over time
Comprehensive Documentation: Include all relevant details in reports to provide full context
Consistent Branding: Use custom templates that match organizational branding
Include Visualizations: Use charts and graphs to make complex data more accessible
Multiple Formats: Generate both HTML (for internal review) and PDF (for formal documentation)
Secure Storage: Store reports securely and implement appropriate access controls
Actionable Insights: Focus on providing clear recommendations for addressing issues
Integration with Workflows: Automate report generation as part of regular workflows
Stakeholder Focus: Tailor reports to the needs of different stakeholders (legal, technical, executive)