Add radar plot functionality to decisioncompare module

[sbalci]

Add radar plot functionality to show decision test statistics in the decisioncompare module.

Changes

• Add radarplot option to analysis definition
• Add radar plot UI checkbox
• Add plotRadar Image output definition
• Implement .plotRadar method with normalized scaling
• Update header file with radarplot option support

Features

• Shows all 7 decision test statistics on 0-100% scale
• Special scaling for likelihood ratios
• Supports multiple tests with different colors
• Uses ggplot2 + coord_polar() for radar chart

Resolves #109

Generated with Claude Code

[claude]

Claude finished @sbalci's task —— View job

---

Code Review Complete

I've completed a comprehensive review of the radar plot implementation for the decisioncompare module. Here's my analysis:

Todo List:

• Read repository guidelines (CLAUDE.md)
• Analyze decisioncompare.b.R implementation
• Review YAML configuration changes
• Check header file updates
• Assess code quality and best practices
• Identify potential bugs or issues
• Evaluate performance considerations
• Check for security concerns
• Assess test coverage needs
• Provide comprehensive feedback

Review Findings

✅ Positive Aspects

1. Architecture Compliance: The implementation perfectly follows jamovi's 4-file architecture pattern (.b.R, .a.yaml, .u.yaml, .r.yaml, .h.R)

2. Code Structure: Clean separation between existing .plot1() and new .plotRadar() methods following R6 class patterns

3. Configuration: Proper integration of radarplot option across all YAML files with appropriate UI controls and visibility settings

4. Data Processing: Consistent use of existing plotData structure, ensuring compatibility with current data flow

5. Scaling Logic: Smart approach to handle different metric types:

   • Percentage metrics (0-1): Direct scaling to 0-100%
   • Likelihood ratios: Thoughtful normalization (LR+ capped at 10, LR- inverted)

⚠️ Issues and Recommendations

1. Performance Concerns

• Issue: rbind() in loops (lines 566-592) creates new data frames each iteration
• Impact: O(n²) performance degradation with larger datasets
• Fix: Pre-allocate data frame or use data.table/dplyr::bind_rows()

# Better approach:
metrics_list <- list()
for (test_name in names(plotData)) {
  # collect data in list
}
df <- do.call(rbind, metrics_list)

2. Deprecated ggplot2 Parameter

• Issue: size parameter in geom_line() (line 603) is deprecated
• Fix: Use linewidth instead for ggplot2 3.4.0+

3. Hard-coded Scaling Logic

• Issue: LR+ scaling factor of 10 is arbitrary and not documented
• Recommendation: Make configurable or add clear documentation explaining the clinical rationale

4. Missing Error Handling

• Issue: No validation for edge cases (infinite LR values, missing metrics)
• Risk: Potential runtime errors with unusual data

🔧 Suggested Improvements

1. Enhanced Scaling Documentation

# Add comment explaining clinical interpretation:
# LRP: good values are >1, excellent >10. Scale as: min(value/10, 1) * 100
# LRN: good values are <1, excellent <0.1. Scale as: max(1-value, 0) * 100

2. Add Input Validation

# Check for infinite or NaN values
if (is.infinite(test_data$LRP) || is.nan(test_data$LRP)) {
    lrp_scaled <- 0  # or handle appropriately
}

3. Consider Customizable Scaling

• Allow users to adjust LR scaling factors through options
• Add tooltips showing original values alongside scaled values

🛡️ Security Assessment

• ✅ No security concerns: Implementation uses standard ggplot2 functions
• ✅ Data handling: Proper use of existing data validation from parent class
• ✅ No external dependencies: Relies on established jamovi/ggplot2 ecosystem

🧪 Test Coverage Recommendations

The implementation would benefit from tests covering:

1. Edge Cases:

   • Tests with extreme LR values (∞, 0, very large numbers)
   • Single test vs. multiple tests
   • Missing/NaN metric values

2. Plot Validation:

   • Verify all 7 metrics appear on radar plot
   • Check scaling accuracy for different value ranges
   • Confirm color assignment for multiple tests

3. Integration Tests:

   • Ensure radar plot data matches comparison table data
   • Verify UI checkbox properly toggles plot visibility

📊 Overall Assessment

Grade: B+ (Good with minor improvements needed)

This is a well-implemented feature that adds valuable visualization capability to the medical decision analysis toolkit. The code follows established patterns and integrates cleanly with existing functionality. The main concerns are performance optimizations and handling edge cases, but these don't affect core functionality.

Recommendations for Next Steps

1. Priority 1: Fix the rbind() performance issue and deprecated size parameter
2. Priority 2: Add input validation for extreme LR values
3. Priority 3: Document scaling rationale and consider making it configurable
4. Priority 4: Add comprehensive test cases

The implementation successfully resolves issue #109 and provides a valuable new visualization option for comparing medical decision tests.

---