Visualize evaluation metrics and identify patterns

[kctong529]

Goal

Make route-pattern and failure-mode results easier to interpret.

Description

Create visual summaries of evaluated route-navigation results.

The goal is not just to make charts, but to use visualisation to identify meaningful patterns in model behavior.

Useful visual summaries may include:

• success/failure by route pattern
• failure-mode frequency
• prompt-template comparison
• GPT vs Gemini comparison
• route property vs performance
• valid path vs exact shortest path
• relative route-length error
• examples of solved, partly solved, and failed routes

This issue should support the Sprint 4 result interpretation and final presentation/report.

Suggested work

• Decide which metrics are available from saved evaluation results.
• Create at least two useful visual summaries.
• Use route-pattern labels where available.
• Use failure-mode labels where available.
• Compare GPT/Gemini or prompt templates if enough data exists.
• Save useful figures/screenshots for report or presentation use.
• Add short notes explaining the main takeaway of each figure.

Acceptance criteria

• At least two useful visual summaries are produced.
• Visualisations use actual saved evaluation results.
• Figures support interpretation rather than decoration.
• Captions or notes explain the main takeaway.
• Limitations of the small project scale are noted.
• Outputs can be used in the final report or presentation.