feat(plotnine): implement lift-curve (#2389)

## Implementation: `lift-curve` - plotnine

Implements the **plotnine** version of `lift-curve`.

**File:** `plots/lift-curve/implementations/plotnine.py`

---
:robot: *[impl-generate
workflow](https://github.com/MarkusNeusinger/pyplots/actions/runs/20543282476)*

---------

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

Author: github-actions[bot]

plots/lift-curve/implementations/plotnine.py

@@ -0,0 +1,102 @@
+""" pyplots.ai
+lift-curve: Model Lift Chart
+Library: plotnine 0.15.2 | Python 3.13.11
+Quality: 92/100 | Created: 2025-12-27
+"""
+
+import numpy as np
+import pandas as pd
+from plotnine import (
+    aes,
+    element_text,
+    geom_hline,
+    geom_line,
+    geom_point,
+    ggplot,
+    labs,
+    scale_x_continuous,
+    scale_y_continuous,
+    theme,
+    theme_minimal,
+)
+
+
+# Data - simulated customer response model scores
+np.random.seed(42)
+n_samples = 1000
+
+# Generate realistic response probabilities
+# Assume a model that has learned some signal
+base_prob = 0.15  # 15% baseline response rate
+model_score = np.random.beta(2, 5, n_samples)  # Model predictions
+
+# True responses correlated with model score (good model)
+response_prob = 0.05 + 0.6 * model_score  # Higher score = higher response chance
+y_true = (np.random.random(n_samples) < response_prob).astype(int)
+y_score = model_score + np.random.normal(0, 0.05, n_samples)  # Add noise
+y_score = np.clip(y_score, 0, 1)
+
+# Calculate lift curve data
+# Sort by predicted score descending
+sorted_indices = np.argsort(y_score)[::-1]
+y_true_sorted = y_true[sorted_indices]
+
+# Calculate cumulative lift
+n_total = len(y_true)
+n_positive = y_true.sum()
+baseline_rate = n_positive / n_total
+
+# Calculate cumulative values at each decile percentage
+percentiles = np.arange(1, 101)
+lift_values = []
+pct_population = []
+
+for pct in percentiles:
+    n_targeted = int(np.ceil(n_total * pct / 100))
+    n_positive_captured = y_true_sorted[:n_targeted].sum()
+
+    # Lift = (response rate in targeted group) / (baseline response rate)
+    targeted_rate = n_positive_captured / n_targeted
+    lift = targeted_rate / baseline_rate if baseline_rate > 0 else 0
+
+    lift_values.append(lift)
+    pct_population.append(pct)
+
+# Create DataFrame for plotting
+df = pd.DataFrame({"pct_population": pct_population, "lift": lift_values})
+
+# Add reference line data
+df_reference = pd.DataFrame({"pct_population": [0, 100], "lift": [1.0, 1.0]})
+
+# Create key points for markers (at deciles)
+decile_points = df[df["pct_population"].isin([10, 20, 30, 40, 50, 60, 70, 80, 90, 100])]
+
+# Plot
+plot = (
+    ggplot()
+    + geom_hline(yintercept=1.0, linetype="dashed", color="#888888", size=1.2, alpha=0.7)
+    + geom_line(data=df, mapping=aes(x="pct_population", y="lift"), color="#306998", size=2.5)
+    + geom_point(
+        data=decile_points,
+        mapping=aes(x="pct_population", y="lift"),
+        color="#306998",
+        size=5,
+        fill="#FFD43B",
+        stroke=1.5,
+    )
+    + scale_x_continuous(breaks=[0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100], limits=(0, 100))
+    + scale_y_continuous(breaks=[0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5], limits=(0, None))
+    + labs(title="lift-curve · plotnine · pyplots.ai", x="Population Targeted (%)", y="Cumulative Lift")
+    + theme_minimal()
+    + theme(
+        figure_size=(16, 9),
+        text=element_text(size=14),
+        axis_title=element_text(size=20),
+        axis_text=element_text(size=16),
+        plot_title=element_text(size=24),
+        panel_grid_minor=element_text(alpha=0),
+    )
+)
+
+# Save
+plot.save("plot.png", dpi=300, verbose=False)

plots/lift-curve/metadata/plotnine.yaml

@@ -0,0 +1,24 @@
+library: plotnine
+specification_id: lift-curve
+created: '2025-12-27T19:20:56Z'
+updated: '2025-12-27T19:26:28Z'
+generated_by: claude-opus-4-5-20251101
+workflow_run: 20543282476
+issue: 0
+python_version: 3.13.11
+library_version: 0.15.2
+preview_url: https://storage.googleapis.com/pyplots-images/plots/lift-curve/plotnine/plot.png
+preview_thumb: https://storage.googleapis.com/pyplots-images/plots/lift-curve/plotnine/plot_thumb.png
+preview_html: null
+quality_score: 92
+review:
+  strengths:
+  - Excellent implementation of lift curve with correct mathematical calculation
+  - Clean use of plotnine grammar of graphics (ggplot, geom_line, geom_point, geom_hline)
+  - Good visual distinction with yellow-filled decile markers on blue line
+  - Proper reference line at y=1 for random selection baseline
+  - Well-documented code with realistic customer response scenario
+  - Appropriate sizing for high-resolution output (16x9, 300 DPI)
+  weaknesses:
+  - Y-axis label could include (ratio) for clarity since lift is a unitless ratio
+  - Could add text annotations at key decile points showing actual lift values