Examples

This section provides comprehensive examples showing how to use optimal-classification-cutoffs in various scenarios. All examples are executable Jupyter notebooks that demonstrate the power of API 2.0.0.

Note

All notebooks are executed automatically during documentation build to ensure they work with the latest code. You can also download and run them locally.

Interactive Notebooks

Notebook Examples

Learning Path

For the best learning experience, follow this order:

  1. Quickstart → See immediate 40%+ improvements with minimal code

  2. Business Value → Understand cost-sensitive optimization for real ROI

  3. Multiclass → Master advanced multiclass threshold strategies

  4. Interactive Demo → Explore mathematical foundations interactively

Each example builds on the previous ones while being self-contained enough to run independently.

Running the Examples

To run these examples locally:

  1. Install with examples dependencies:

    pip install optimal-classification-cutoffs[examples]

  2. Download the notebooks from the GitHub repository

  3. Launch Jupyter:

    jupyter notebook

  4. Open and run any of the example notebooks

What You’ll Learn

API 2.0.0 Features Demonstrated:

  • Progressive disclosure design (simple → advanced)

  • Explainable auto-selection with reasoning

  • Enum-based explicit control (Task, Average)

  • Namespace organization (metrics/, cv/, bayes/, algorithms/)

  • Modern match/case performance optimizations

  • API v2.0 with modernized design patterns

Real-World Applications:

  • Fraud detection with cost-sensitive optimization

  • Medical diagnosis with asymmetric error costs

  • Document classification with multiclass strategies

  • A/B testing with threshold validation

  • Business ROI calculation from model improvements

Each notebook contains working code you can run immediately to see 40%+ performance improvements over default 0.5 thresholds.