Matplotlib

Basics: Common Plot Types
Subplots & Advanced Plots
Example project

Matplotlib permet crear tot tipus de gràfics en Python: línies, barres, pastís, dispersió, histogrames i més. Primer generes les dades (normalment amb NumPy), després “construeixes” el gràfic afegint elements com etiquetes, títols, llegendes i quadrícules. Finalment, plt.show() el mostra. Cada comanda que dibuixa dades afegeix un nou element visual, i pots personalitzar-los amb colors, tipus de línia, formes i altres estils. També és possible crear diverses figures i subplots per mostrar diferents gràfics dins del mateix espai, cosa que facilita comparar dades i visualitzacions.

Basics: Common Plot Types

Visualize data using line plots, scatter plots, histograms, bar charts, and pie charts.

import numpy as np
import matplotlib.pyplot as plt
np.random.seed(3)

Line Plot

Use plt.plot() for continuous data.
Add labels, title, grid, and legend.

x = np.linspace(0, 2 * np.pi, 200)
y1 = np.sin(x)
y2 = np.cos(x)

plt.plot(x, y1, label='sin(x)')
plt.plot(x, y2, label='cos(x)', linestyle='--')
plt.xlabel('x')
plt.ylabel('f(x)')
plt.title('Sine and Cosine Waves')
plt.legend()
plt.grid(True)
plt.show()

png

Scatter Plot

plt.scatter() for showing individual data points.
Color and size can represent additional dimensions.

x = np.random.rand(100)
y = np.random.rand(100)
colors = np.random.rand(100)
sizes = 100 * np.random.rand(100)

plt.scatter(x, y, c=colors, s=sizes, alpha=0.6, cmap='viridis')
plt.colorbar(label='Color scale')
plt.title('Bubble Scatter Plot')
plt.xlabel('x')
plt.ylabel('y')
plt.show()

png

Histogram

plt.hist() to visualize distributions.
Adjust bins and density.

data = np.random.randn(1000)
plt.hist(data, bins=30, density=True, alpha=0.7)
plt.title('Normalized Histogram of Gaussian Data')
plt.xlabel('Value')
plt.ylabel('Density')
plt.show()

png

Bar Chart

plt.bar() for categorical comparisons.
Horizontal bar: plt.barh()

categories = ['A', 'B', 'C', 'D']
values = [5, 7, 3, 4]

plt.bar(categories, values, alpha=0.8)
plt.title('Vertical Bar Chart')
plt.xlabel('Category')
plt.ylabel('Value')
plt.show()

# Horizontal bar chart
plt.barh(categories, values, alpha=0.8)
plt.title('Horizontal Bar Chart')
plt.xlabel('Value')
plt.ylabel('Category')
plt.show()

png

Pie Chart

plt.pie() for composition of a whole.

labels = ['W', 'X', 'Y', 'Z']
sizes = [15, 30, 45, 10]
explode = (0, 0.1, 0, 0)  # only 'explode' the 2nd slice

plt.pie(sizes, labels=labels, autopct='%1.1f%%', explode=explode, shadow=True)
plt.title('Pie Chart Example')
plt.axis('equal')  # Equal aspect ensures pie is drawn as a circle.
plt.show()

png

Saving Figures

Use plt.savefig() to export plots.

# Example: Save the sine wave plot
x = np.linspace(0, 2 * np.pi, 200)  # Redefine x for the sine wave
y1 = np.sin(x)  # Redefine y1 based on the correct x

plt.figure()
plt.plot(x, y1)
plt.title('Sine Wave')
plt.savefig('data/sine_wave.png', dpi=150)
print('Saved figure as sine_wave.png')

Saved figure as sine_wave.png

png

Subplots & Advanced Plots

Learn to create complex layouts and advanced visualizations such as error bars and images.

import numpy as np
import matplotlib.pyplot as plt
np.random.seed(4)

Creating Subplots

plt.subplots(nrows, ncols) to create grid of axes.
Adjust layouts with figsize, tight_layout.

fig, axs = plt.subplots(2, 2, figsize=(10, 8))
axs[0, 0].plot(np.sin(np.linspace(0, 2*np.pi, 100)))
axs[0, 0].set_title('Sine Wave')

axs[0, 1].bar(['A', 'B', 'C'], [3, 5, 2])
axs[0, 1].set_title('Bar Plot')

axs[1, 0].hist(np.random.randn(500), bins=20)
axs[1, 0].set_title('Histogram')

axs[1, 1].scatter(np.random.rand(50), np.random.rand(50), c=np.random.rand(50), cmap='plasma')
axs[1, 1].set_title('Scatter Plot')

plt.tight_layout()
plt.show()

png

Shared Axes and Figure-Level Settings

Share x or y axes across subplots.
Add a main title with fig.suptitle().

fig, axs = plt.subplots(2, 1, sharex=True, figsize=(8, 6))
x = np.linspace(0, 10, 100)
axs[0].plot(x, np.sin(x))
axs[0].set_title('Sine')
axs[1].plot(x, np.cos(x))
axs[1].set_title('Cosine')
fig.suptitle('Shared X-Axis Example')
plt.show()

png

Error Bars

Use plt.errorbar() to show uncertainties.

x = np.arange(5)
y = np.random.rand(5)
yerr = 0.1 + 0.2 * np.random.rand(5)
plt.errorbar(x, y, yerr=yerr, fmt='o-', capsize=5)
plt.title('Error Bars Example')
plt.xlabel('X')
plt.ylabel('Y')
plt.show()

png

Image Plot

Display 2D arrays as images with plt.imshow().

img = np.random.rand(10, 10)
plt.imshow(img, interpolation='nearest')
plt.title('Random Image')
plt.colorbar()
plt.show()

png

Advanced: Contour and Heatmap

Use plt.contour() and plt.imshow() for heatmaps.

x = np.linspace(-3, 3, 100)
y = np.linspace(-3, 3, 100)
X, Y = np.meshgrid(x, y)
Z = np.exp(-(X**2 + Y**2))

plt.figure(figsize=(6,5))
contours = plt.contour(X, Y, Z, levels=6)
plt.clabel(contours, inline=True)
plt.title('Contour Plot')
plt.show()

# Heatmap
plt.figure(figsize=(6,5))
plt.imshow(Z, origin='lower', extent=[-3,3,-3,3])
plt.title('Heatmap')
plt.colorbar()
plt.show()

png

Example project

This section guides you through generating, analyzing, and visualizing a synthetic dataset for classification.

import numpy as np
import matplotlib.pyplot as plt
from sklearn.datasets import make_classification
np.random.seed(5)

Generate Synthetic Data

Use make_classification to create a dataset.

X, y = make_classification(
    n_samples=200,
    n_features=4,
    n_informative=2,
    n_redundant=0,
    n_clusters_per_class=1,
    flip_y=0.01,
    class_sep=1.5,
    random_state=5
)
print('Features shape:', X.shape)
print('Labels distribution:', np.bincount(y))

Features shape: (200, 4)
Labels distribution: [ 99 101]

Explore Feature Distributions

Plot histograms for each feature.

fig, axs = plt.subplots(2, 2, figsize=(10, 8))
for i in range(4):
    ax = axs.flat[i]
    ax.hist(X[:, i], bins=20, color='C{}'.format(i), alpha=0.7)
    ax.set_title(f'Feature {i} Distribution')
plt.tight_layout()
plt.show()

png

Pairwise Scatter Plots

Visualize relationships between pairs of features.

fig, axs = plt.subplots(2, 3, figsize=(12, 8))
pairs = [(0, 1), (0, 2), (0, 3), (1, 2), (1, 3), (2, 3)]
for ax, (i, j) in zip(axs.flat, pairs):
    ax.scatter(X[:, i], X[:, j], c=y, cmap='coolwarm', alpha=0.6)
    ax.set_xlabel(f'Feature {i}')
    ax.set_ylabel(f'Feature {j}')
    ax.set_title(f'{i} vs {j}')
plt.tight_layout()
plt.show()

png

Compute Basic Statistics

Calculate means and standard deviations per class.

classes = np.unique(y)
for cls in classes:
    cls_data = X[y == cls]
    print(f'Class {cls}: mean={cls_data.mean(axis=0)}, std={cls_data.std(axis=0)}')

Class 0: mean=[-1.51904518 -0.13961004 -1.51551924 -0.10388006], std=[0.94565383 0.92078405 0.53456453 0.88192832]
Class 1: mean=[-1.62540423  0.03701214  1.51590761  0.1083452 ], std=[0.85187348 0.96153734 0.53164293 0.91483422]

Train/Test Split & Visualization

Split data using NumPy and plot class distributions.

indices = np.arange(len(y))
np.random.shuffle(indices)
split = int(0.8 * len(indices))
train_idx, test_idx = indices[:split], indices[split:]
X_train, X_test = X[train_idx], X[test_idx]
y_train, y_test = y[train_idx], y[test_idx]

# Visualize train vs test in feature 0 & 1
plt.figure(figsize=(6,5))
plt.scatter(X_train[:,0], X_train[:,1], c=y_train, cmap='viridis', label='Train', alpha=0.6)
plt.scatter(X_test[:,0], X_test[:,1], c=y_test, cmap='coolwarm', marker='x', label='Test', alpha=0.6)
plt.xlabel('Feature 0')
plt.ylabel('Feature 1')
plt.legend()
plt.title('Train vs Test Data')
plt.show()

png

Save Processed Data

Optionally save arrays to disk for later use.

np.save('data/X_train.npy', X_train)
np.save('data/X_test.npy', X_test)
np.save('data/y_train.npy', y_train)
np.save('data/y_test.npy', y_test)
print('Saved .npy files in data directory.')

Saved .npy files in data directory.

Keyboard shortcuts

apunts.jg5.dev