Agent♥︎Age
Catalog

Calibration Explorer

Official

by m-sanchez · TypeScript

Check classifier confidence against accuracy, fit temperature scaling and save local reports.

io.github.m-sanchez/calibration-explorer (MCP Server)

The MCP server “io.github.m-sanchez/calibration-explorer” provides functionality to check classifier confidence against accuracy, perform fit temperature scaling, and save local reports based on those results.

🛠️ Key Features

  • Check classifier confidence against accuracy
  • Fit temperature scaling
  • Save local reports

🚀 Use Cases

  • Evaluating how classifier confidence aligns with accuracy
  • Applying temperature scaling to calibration
  • Generating and storing local calibration reports

⚡ Developer Benefits

  • Supports calibration analysis using confidence and accuracy comparisons
  • Enables temperature scaling workflows
  • Produces local report outputs for later review

⚠️ Limitations

  • No additional capabilities, tool list, or operational details were provided beyond confidence-vs-accuracy checking, temperature scaling fitting, and local report saving.

Topics

calibrationdata-visualizationmachine-learningmodel-evaluationtypescriptclassifierclaude-codeexpected-calibration-errormcpmcp-serverreliability-diagramtemperature-scaling