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.
""" anyplot.ai
precision-recall: Precision-Recall Curve
Library: matplotlib 3.10.9 | Python 3.13.13
Quality: 81/100 | Updated: 2026-05-10
"""
import matplotlib.pyplot as plt
import numpy as np
# Generate synthetic classification data with imbalanced classes
np.random.seed(42)
n_samples = 500
# Simulate imbalanced dataset: 20% positive, 80% negative
positive_ratio = 0.2
n_positive = int(n_samples * positive_ratio)
n_negative = n_samples - n_positive
# True labels
y_true = np.concatenate([np.ones(n_positive), np.zeros(n_negative)])
# Simulate classifier scores - good classifier gives higher scores to positives
positive_scores = np.random.beta(5, 2, n_positive) # Skewed higher
negative_scores = np.random.beta(2, 5, n_negative) # Skewed lower
y_scores = np.concatenate([positive_scores, negative_scores])
# Calculate precision-recall curve
# Sort by scores descending
desc_score_indices = np.argsort(y_scores)[::-1]
y_scores_sorted = y_scores[desc_score_indices]
y_true_sorted = y_true[desc_score_indices]
# Get unique thresholds
distinct_value_indices = np.where(np.diff(y_scores_sorted))[0]
threshold_idxs = np.concatenate([[0], distinct_value_indices + 1])
# Calculate TP, FP cumulative sums
tps = np.cumsum(y_true_sorted)
fps = np.cumsum(1 - y_true_sorted)
# Calculate precision and recall at each threshold
precision_vals = tps / (tps + fps)
recall_vals = tps / tps[-1]
# Use thresholds at distinct values (add starting point: recall=0, precision=1)
precision = np.concatenate([[1], precision_vals[threshold_idxs]])
recall = np.concatenate([[0], recall_vals[threshold_idxs]])
# Calculate Average Precision (area under curve)
recall_diff = np.diff(recall)
average_precision = np.sum(precision[1:] * recall_diff)
# Create plot
fig, ax = plt.subplots(figsize=(16, 9))
# Plot precision-recall curve with stepped line style
ax.step(
recall, precision, where="post", color="#306998", linewidth=3, label=f"Classifier (AP = {average_precision:.3f})"
)
# Fill area under curve for visual emphasis
ax.fill_between(recall, precision, step="post", alpha=0.2, color="#306998")
# Baseline: random classifier (horizontal line at positive class ratio)
ax.axhline(
y=positive_ratio,
color="#FFD43B",
linestyle="--",
linewidth=2.5,
label=f"Random Baseline (P = {positive_ratio:.0%})",
)
# Iso-F1 curves
f1_scores = np.linspace(0.2, 0.8, num=4)
for f1_score in f1_scores:
x = np.linspace(0.01, 1, 100)
y = f1_score * x / (2 * x - f1_score)
mask = (y >= 0) & (y <= 1) & (x >= f1_score / 2)
ax.plot(x[mask], y[mask], color="gray", alpha=0.3, linewidth=1.5, linestyle=":")
# Add F1 label at end of curve
if np.any(mask):
label_x = x[mask][-1]
label_y = y[mask][-1]
ax.annotate(
f"F1={f1_score:.1f}", xy=(label_x, label_y), fontsize=12, color="gray", alpha=0.7, ha="left", va="bottom"
)
# Styling
ax.set_xlabel("Recall", fontsize=20)
ax.set_ylabel("Precision", fontsize=20)
ax.set_title("precision-recall · matplotlib · pyplots.ai", fontsize=24)
ax.tick_params(axis="both", labelsize=16)
ax.set_xlim([0.0, 1.0])
ax.set_ylim([0.0, 1.05])
ax.legend(loc="upper right", fontsize=16)
ax.grid(True, alpha=0.3, linestyle="--")
plt.tight_layout()
plt.savefig("plot.png", dpi=300, bbox_inches="tight")
Part of Precision-Recall Curve on anyplot.ai.