LeadIndicator

A Python package for analyzing leading indicators and finding optimal thresholds.

Features

• Analyze time series data to identify leading indicators
• Find optimal thresholds using multiple metrics
• Generate comprehensive analysis reports
• Support for categorical analysis
• Beautiful visualizations
• Polars-based for high performance

Installation

pip install leadindicator

Quick Start

import polars as pl
from leadindicator import ThresholdAnalyzer

# Load your data
df = pl.read_excel("your_data.xlsx")

# Create analyzer
analyzer = ThresholdAnalyzer(
    score_column="score",
    target_column="event",
    category_column="category"  # Optional
)

# Run analysis
results = analyzer.analyze(df)

# Generate report
analyzer.save_report("analysis_report.html")

Example: Fruit Spoilage Analysis

The package includes an example that demonstrates how to use LeadIndicator to analyze fruit spoilage data:

from leadindicator import ThresholdAnalyzer
import polars as pl

# Load example data
df = pl.read_excel("Test Data.xlsx")

# Create analyzer
analyzer = ThresholdAnalyzer(
    score_column="Ethylene",
    target_column="Spoiled",
    category_column="Fruit Type"
)

# Run analysis
results = analyzer.analyze(df)

# Save report
analyzer.save_report("fruit_spoilage_analysis.html")

# Access results programmatically
best_threshold = results[0]
print(f"Best threshold: {best_threshold.threshold}")
print(f"Balanced accuracy: {best_threshold.balanced_accuracy:.2%}")
print(f"Capture rate: {best_threshold.capture_rate:.1f}%")
print(f"Event probability: {best_threshold.event_probability:.1f}%")

Documentation

For detailed documentation, visit docs/.

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

License

This project is licensed under the MIT License - see the LICENSE file for details.