lawkit-python 2.1.1

Python wrapper for lawkit - Statistical law analysis toolkit for fraud detection and data quality assessment

lawkit-python

Python wrapper for the lawkit CLI tool - Statistical law analysis toolkit for fraud detection and data quality assessment.

Installation

pip install lawkit-python

This will automatically download the appropriate lawkit binary for your system from GitHub Releases.

Quick Start

import lawkit

# Analyze financial data with Benford Law
result = lawkit.analyze_benford('financial_data.csv')
print(result)

# Get structured JSON output
json_result = lawkit.analyze_benford(
    'accounting.csv',
    lawkit.LawkitOptions(format='csv', output='json')
)
print(f"Risk level: {json_result.risk_level}")
print(f"P-value: {json_result.p_value}")

# Check if data follows Pareto principle (80/20 rule)
pareto_result = lawkit.analyze_pareto(
    'sales_data.csv',
    lawkit.LawkitOptions(output='json', gini_coefficient=True)
)
print(f"Gini coefficient: {pareto_result.gini_coefficient}")
print(f"80/20 concentration: {pareto_result.concentration_80_20}")

Features

Statistical Laws Supported

• Benford Law: Detect fraud and anomalies in numerical data
• Pareto Principle: Analyze 80/20 distributions and concentration
• Zipf Law: Analyze word frequencies and power-law distributions
• Normal Distribution: Test for normality and detect outliers
• Poisson Distribution: Analyze rare events and count data

Advanced Analysis

• Multi-law Comparison: Compare multiple statistical laws on the same data
• Outlier Detection: Advanced anomaly detection algorithms
• Time Series Analysis: Trend and seasonality detection
• International Numbers: Support for various number formats (Japanese, Chinese, etc.)
• Memory Efficient: Handle large datasets with streaming analysis

File Format Support

• CSV, JSON, YAML, TOML, XML: Standard structured data formats
• Excel Files: .xlsx and .xls support
• PDF Documents: Extract and analyze numerical data from PDFs
• Word Documents: Analyze data from .docx files
• PowerPoint: Extract data from presentations

Usage Examples

Modern API (Recommended)

import lawkit

# Analyze with Benford Law
result = lawkit.analyze_benford('invoice_data.csv')
print(result)

# Get detailed JSON analysis
json_result = lawkit.analyze_benford(
    'financial_statements.xlsx',
    lawkit.LawkitOptions(
        format='excel',
        output='json',
        confidence=0.95,
        verbose=True
    )
)

if json_result.risk_level == "High":
    print("⚠️  High risk of fraud detected!")
    print(f"Chi-square: {json_result.chi_square}")
    print(f"P-value: {json_result.p_value}")
    print(f"MAD: {json_result.mad}%")

# Pareto analysis for business insights
pareto_result = lawkit.analyze_pareto(
    'customer_revenue.csv',
    lawkit.LawkitOptions(
        output='json',
        gini_coefficient=True,
        business_analysis=True,
        percentiles="70,80,90"
    )
)

print(f"Top 20% customers generate {pareto_result.concentration_80_20:.1f}% of revenue")
print(f"Income inequality (Gini): {pareto_result.gini_coefficient:.3f}")

# Normal distribution analysis with outlier detection
normal_result = lawkit.analyze_normal(
    'quality_measurements.csv',
    lawkit.LawkitOptions(
        output='json',
        outlier_detection=True,
        test_type='shapiro'
    )
)

if normal_result.p_value < 0.05:
    print("Data does not follow normal distribution")
    if normal_result.outliers:
        print(f"Found {len(normal_result.outliers)} outliers")

# Multi-law comparison
comparison = lawkit.compare_laws(
    'complex_dataset.csv',
    lawkit.LawkitOptions(output='json')
)
print(f"Best fitting law: {comparison.data.get('best_law')}")
print(f"Overall risk level: {comparison.risk_level}")

Generate Sample Data

import lawkit

# Generate Benford Law compliant data
benford_data = lawkit.generate_data('benf', samples=1000, seed=42)
print(benford_data)

# Generate normal distribution data
normal_data = lawkit.generate_data('normal', samples=500, mean=100, stddev=15)

# Generate Pareto distribution data
pareto_data = lawkit.generate_data('pareto', samples=1000, concentration=0.8)

# Test the pipeline: generate → analyze
data = lawkit.generate_data('benf', samples=10000, seed=42)
result = lawkit.analyze_string(data, 'benf', lawkit.LawkitOptions(output='json'))
print(f"Generated data risk level: {result.risk_level}")

Analyze String Data Directly

import lawkit

# Analyze CSV data from string
csv_data = """amount
123.45
456.78
789.12
234.56
567.89"""

result = lawkit.analyze_string(
    csv_data,
    'benf',
    lawkit.LawkitOptions(format='csv', output='json')
)
print(f"Risk assessment: {result.risk_level}")

# Analyze JSON data
json_data = '{"values": [12, 23, 34, 45, 56, 67, 78, 89]}'
result = lawkit.analyze_string(
    json_data,
    'normal',
    lawkit.LawkitOptions(format='json', output='json')
)
print(f"Is normal: {result.p_value > 0.05}")

Advanced Options

import lawkit

# High-performance analysis with optimization
result = lawkit.analyze_benford(
    'large_dataset.csv',
    lawkit.LawkitOptions(
        optimize=True,
        parallel=True,
        memory_efficient=True,
        min_count=50,
        threshold=0.001
    )
)

# International number support
result = lawkit.analyze_benford(
    'japanese_accounting.csv',
    lawkit.LawkitOptions(
        international=True,
        format='csv',
        output='json'
    )
)

# Time series analysis
result = lawkit.analyze_normal(
    'sensor_data.csv',
    lawkit.LawkitOptions(
        time_series=True,
        outlier_detection=True,
        output='json'
    )
)

Legacy API (Backward Compatibility)

from lawkit import run_lawkit

# Direct command execution
result = run_lawkit(["benf", "data.csv", "--format", "csv", "--output", "json"])

if result.returncode == 0:
    print("Analysis successful")
    print(result.stdout)
else:
    print("Analysis failed")
    print(result.stderr)

# Legacy analysis functions
from lawkit.compat import run_benford_analysis, run_pareto_analysis

benford_result = run_benford_analysis("financial.csv", format="csv", output="json")
pareto_result = run_pareto_analysis("sales.csv", gini_coefficient=True)

Installation and Setup

Automatic Installation (Recommended)

pip install lawkit-python

The package will automatically download the appropriate binary for your platform.

Manual Binary Installation

If automatic download fails:

lawkit-download-binary

Development Installation

git clone https://github.com/kako-jun/lawkit
cd lawkit/lawkit-python
pip install -e .[dev]

Verify Installation

import lawkit

# Check if lawkit is available
if lawkit.is_lawkit_available():
    print("✅ lawkit is installed and working")
    print(f"Version: {lawkit.get_version()}")
else:
    print("❌ lawkit is not available")

# Run self-test
if lawkit.selftest():
    print("✅ All tests passed")
else:
    print("❌ Self-test failed")

Use Cases

Financial Fraud Detection

import lawkit

# Analyze invoice amounts for fraud
result = lawkit.analyze_benford('invoices.csv', 
                               lawkit.LawkitOptions(output='json'))

if result.risk_level in ['High', 'Critical']:
    print("🚨 Potential fraud detected in invoice data")
    print(f"Statistical significance: p={result.p_value:.6f}")
    print(f"Deviation from Benford Law: {result.mad:.2f}%")

Business Intelligence

import lawkit

# Analyze customer revenue distribution
result = lawkit.analyze_pareto('customer_revenue.csv',
                              lawkit.LawkitOptions(
                                  output='json',
                                  business_analysis=True,
                                  gini_coefficient=True
                              ))

print(f"Revenue concentration: {result.concentration_80_20:.1f}%")
print(f"Market inequality: {result.gini_coefficient:.3f}")

Quality Control

import lawkit

# Analyze manufacturing measurements
result = lawkit.analyze_normal('measurements.csv',
                              lawkit.LawkitOptions(
                                  output='json',
                                  outlier_detection=True,
                                  test_type='shapiro'
                              ))

if result.p_value < 0.05:
    print("⚠️  Process out of control - not following normal distribution")
    if result.outliers:
        print(f"Found {len(result.outliers)} outlying measurements")

Text Analysis

import lawkit

# Analyze word frequency in documents
result = lawkit.analyze_zipf('document.txt',
                            lawkit.LawkitOptions(output='json'))

print(f"Text follows Zipf Law: {result.p_value > 0.05}")
print(f"Power law exponent: {result.exponent:.3f}")

API Reference

Main Functions

• analyze_benford(input_data, options) - Benford Law analysis
• analyze_pareto(input_data, options) - Pareto principle analysis
• analyze_zipf(input_data, options) - Zipf Law analysis
• analyze_normal(input_data, options) - Normal distribution analysis
• analyze_poisson(input_data, options) - Poisson distribution analysis
• compare_laws(input_data, options) - Multi-law comparison
• generate_data(law_type, samples, **kwargs) - Generate sample data
• analyze_string(content, law_type, options) - Analyze string data directly

Utility Functions

• is_lawkit_available() - Check if lawkit CLI is available
• get_version() - Get lawkit version
• selftest() - Run self-test

Classes

• LawkitOptions - Configuration options for analysis
• LawkitResult - Analysis results with structured access
• LawkitError - Exception class for lawkit errors

Platform Support

• Windows: x86_64
• macOS: x86_64, ARM64 (Apple Silicon)
• Linux: x86_64, ARM64

Requirements

• Python 3.8+
• No additional dependencies required

License

This project is licensed under the MIT License.

Support

• GitHub Issues: https://github.com/kako-jun/lawkit/issues
• Documentation: https://github.com/kako-jun/lawkit/tree/main/docs
• Examples: https://github.com/kako-jun/lawkit/tree/main/docs/user-guide/examples.md

Contributing

Contributions are welcome! Please read the Contributing Guide for details.