[precision-recall] Precision-Recall Curve

[MarkusNeusinger]

Plot Request

Precision-Recall curve for imbalanced classification problems.

Use Cases

• Imbalanced dataset evaluation
• Information retrieval systems
• Medical diagnosis (high recall important)

Expected Features

• Precision vs Recall curve
• Average Precision (AP) score displayed
• Baseline reference (random classifier)
• Optional: multiple curves for comparison

[claude]

✅ Spec ID: precision-recall

Rationale: Precision-Recall curves are a standard visualization for evaluating binary classifiers, especially useful for imbalanced datasets. No existing spec covers this ML evaluation plot type.

Add label approved to begin generation.

[github-actions]

✅ Specification Ready: precision-recall

PR: #2279

---

specification.md

precision-recall: Precision-Recall Curve

Description

A Precision-Recall curve plots precision (positive predictive value) against recall (sensitivity) at various classification thresholds. This visualization is essential for evaluating binary classifiers on imbalanced datasets where accuracy alone is misleading. The area under the curve (Average Precision) summarizes classifier performance, with higher values indicating better performance.

Applications

• Evaluating fraud detection models where fraudulent transactions are rare compared to legitimate ones
• Assessing medical diagnostic systems where correctly identifying positive cases (high recall) is critical
• Comparing information retrieval systems for document search relevance ranking
• Optimizing spam filters to balance catching spam (recall) with minimizing false positives (precision)

Data

• y_true (binary array) - Ground truth binary labels (0 or 1)
• y_scores (numeric array) - Predicted probabilities or decision function scores from classifier
• Size: 100-10000 samples typical for evaluation
• Example: Binary classification predictions from sklearn classifier with predict_proba() output

Notes

• Display Average Precision (AP) score in legend or annotation
• Include baseline reference line showing random classifier performance (horizontal line at positive class ratio)
• Use stepped line style to accurately represent threshold-based curve
• Consider showing iso-F1 curves as contour lines for F1 score reference
• For multiple classifiers comparison, use distinct colors with clear legend

---

Next: Add approved label to merge this specification to main.

---

🤖 spec-create workflow

[github-actions]

🚀 Specification Merged!

Specification: precision-recall
PR: #2279 (merged)

The specification is now in main and synced to PostgreSQL.

---

Next: Add generate:{library} labels to start generating implementations.

Available libraries: matplotlib, seaborn, plotly, bokeh, altair, plotnine, pygal, highcharts, letsplot

Or use workflow dispatch for bulk operations.

---

🤖 spec-create workflow

[MarkusNeusinger]

✅ Closed with 6/9 implementations in repo (precision-recall)