Knee Detection Page

This page documents the Knee Point Detection analysis view in the OptiScope Dash application.

Overview

The Knee Detection page helps users identify the knee points on a Pareto front – the solutions that represent the best trade‑off between competing objectives. Multiple detection methods are available:

• Angle – based on sharp turns in the front.
• Distance – farthest from the line connecting extremes.
• Trade‑off – changes in the trade‑off ratio.
• Curvature – high curvature regions.

Features

• Select an optimization result to analyze.
• Choose a detection method from a dropdown.
• Visualize the first two objectives with knee points highlighted in red stars.
• Supports dark/light themes and responsive layout.

Usage

1. Open the Knee Detection page from the navigation menu.
2. Choose a result using the Result Selector.
3. Pick a detection method from the dropdown.
4. The scatter plot updates, highlighting the detected knee points.

Implementation Details

• Implemented in optiscope/dash_app/pages/knee_detection/page.py.
• Uses optiscope.analysis.knee_detection.detect_knee_points for the core algorithm.
• Follows the same layout conventions as other analysis pages.

Future Enhancements

• Add a region view to display a broader knee region.
• Provide a download option for the identified knee points.