Model Learning Curve — Pygal

A learning curve visualizes model performance (training and validation scores) as a function of training set size. It is essential for diagnosing bias vs variance tradeoffs, determining whether collecting more data would improve model performance, and guiding model selection decisions. The plot typically shows two lines with shaded confidence bands representing variability across cross-validation folds.

Model Learning Curve rendered with Pygal

Python source (Pygal)

""" anyplot.ai
learning-curve-basic: Model Learning Curve
Library: pygal 3.1.0 | Python 3.13.13
Quality: 91/100 | Updated: 2026-05-10
"""

import os

import numpy as np
import pygal
from pygal.style import Style


# Theme tokens
THEME = os.getenv("ANYPLOT_THEME", "light")
PAGE_BG = "#FAF8F1" if THEME == "light" else "#1A1A17"
INK = "#1A1A17" if THEME == "light" else "#F0EFE8"
INK_MUTED = "#6B6A63" if THEME == "light" else "#A8A79F"

IMPRINT = ("#009E73", "#C475FD", "#4467A3", "#BD8233", "#AE3030", "#2ABCCD", "#954477")

# Data - Simulating sklearn's learning_curve output
np.random.seed(42)

# Training set sizes
n_samples_total = 1000
train_sizes_pct = np.array([0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0])
train_sizes = (train_sizes_pct * n_samples_total).astype(int)

# Simulate cross-validation folds (5 folds)
n_folds = 5
n_sizes = len(train_sizes)

# Training scores: start high, remain high (slight overfitting pattern)
train_scores_base = 0.95 - 0.05 * np.exp(-train_sizes / 200)
train_scores_std_vals = 0.02 * np.exp(-train_sizes / 300)
train_scores = np.array(
    [train_scores_base[i] + np.random.randn(n_folds) * train_scores_std_vals[i] for i in range(n_sizes)]
).T

# Validation scores: start lower, converge towards training (gap shows variance)
val_scores_base = 0.65 + 0.25 * (1 - np.exp(-train_sizes / 400))
val_scores_std_vals = 0.04 * np.exp(-train_sizes / 500) + 0.01
val_scores = np.array(
    [val_scores_base[i] + np.random.randn(n_folds) * val_scores_std_vals[i] for i in range(n_sizes)]
).T

# Calculate means and standard deviations
train_mean = np.mean(train_scores, axis=0)
train_std = np.std(train_scores, axis=0)
val_mean = np.mean(val_scores, axis=0)
val_std = np.std(val_scores, axis=0)

# Custom style for 4800x2700 canvas with theme-adaptive colors
custom_style = Style(
    background=PAGE_BG,
    plot_background=PAGE_BG,
    foreground=INK,
    foreground_strong=INK,
    foreground_subtle=INK_MUTED,
    colors=IMPRINT,
    title_font_size=28,
    label_font_size=22,
    major_label_font_size=18,
    legend_font_size=16,
    value_font_size=14,
    stroke_width=4,
)

# Create XY chart for learning curve
chart = pygal.XY(
    width=4800,
    height=2700,
    style=custom_style,
    title="learning-curve-basic · pygal · anyplot.ai",
    x_title="Training Set Size (samples)",
    y_title="Accuracy Score",
    show_dots=True,
    dots_size=8,
    stroke_style={"width": 4},
    show_x_guides=False,
    show_y_guides=True,
    legend_at_bottom=False,
    range=(0.6, 1.02),
    xrange=(50, 1050),
    x_labels=[100, 200, 300, 400, 500, 600, 700, 800, 900, 1000],
    margin=80,
)

# Prepare data points as (x, y) tuples
train_points = [(int(train_sizes[i]), round(train_mean[i], 3)) for i in range(n_sizes)]
val_points = [(int(train_sizes[i]), round(val_mean[i], 3)) for i in range(n_sizes)]

# Add upper/lower bounds for confidence bands (±1 std)
train_upper = [(int(train_sizes[i]), round(train_mean[i] + train_std[i], 3)) for i in range(n_sizes)]
train_lower = [(int(train_sizes[i]), round(train_mean[i] - train_std[i], 3)) for i in range(n_sizes)]
val_upper = [(int(train_sizes[i]), round(val_mean[i] + val_std[i], 3)) for i in range(n_sizes)]
val_lower = [(int(train_sizes[i]), round(val_mean[i] - val_std[i], 3)) for i in range(n_sizes)]

# Add main learning curves
chart.add("Training Score", train_points, stroke_style={"width": 5})
chart.add("Validation Score", val_points, stroke_style={"width": 5})

# Add confidence bounds as secondary lines (thinner, dashed, no legend)
chart.add(None, train_upper, show_dots=False, stroke_style={"width": 2, "dasharray": "8, 4"})
chart.add(None, train_lower, show_dots=False, stroke_style={"width": 2, "dasharray": "8, 4"})
chart.add(None, val_upper, show_dots=False, stroke_style={"width": 2, "dasharray": "8, 4"})
chart.add(None, val_lower, show_dots=False, stroke_style={"width": 2, "dasharray": "8, 4"})

# Save as HTML (interactive) and PNG
chart.render_to_file(f"plot-{THEME}.html")
chart.render_to_png(f"plot-{THEME}.png")

Part of Model Learning Curve on anyplot.ai.

Other implementations