Skip to main content

Data Visualization: Matplotlib & Seaborn

In Machine Learning, a picture is worth a thousand rows of data. Matplotlib is the foundational "grandfather" library for plotting in Python, while Seaborn sits on top of it to provide beautiful, statistically-informed visualizations with much less code.

1. The Anatomy of a Plot​

To master Matplotlib, you must understand its hierarchy. Every plot is contained within a Figure, which can hold one or more Axes (the actual plots).

2. Matplotlib: The Basics​

The most common interface is pyplot. It follows a state-machine logic similar to MATLAB.

import matplotlib.pyplot as plt

# Data
epochs = [1, 2, 3, 4, 5]
loss = [0.9, 0.7, 0.5, 0.3, 0.2]

# Plotting
plt.plot(epochs, loss, label='Training Loss', marker='o')
plt.title("Model Training Progress")
plt.xlabel("Epochs")
plt.ylabel("Loss")
plt.legend()
plt.show()

3. Essential ML Plots​

In your ML workflow, you will constantly use these four types of visualizations:

Plot TypeBest Use CaseLibrary Choice
Line PlotMonitoring Loss/Accuracy over time (epochs).Matplotlib
Scatter PlotFinding correlations between two features ( vs ).Seaborn
HistogramChecking if a feature follows a Normal Distribution.Seaborn
HeatmapVisualizing a Correlation Matrix or Confusion Matrix.Seaborn

4. Seaborn: Statistical Beauty​

Seaborn makes complex plots easy. It integrates directly with Pandas DataFrames and handles the labeling and coloring automatically.

import seaborn as sns

# Load a built-in dataset
iris = sns.load_dataset("iris")

# A single line to see relationships across all features
sns.pairplot(iris, hue="species")
plt.show()

5. Visualizing Model Performance​

The Heatmap (Confusion Matrix)​

A heatmap is the standard way to visualize where a classification model is getting confused.

# Assuming 'cm' is your confusion matrix array
sns.heatmap(cm, annot=True, cmap='Blues')
plt.xlabel("Predicted Label")
plt.ylabel("True Label")

Subplots​

Sometimes you need to compare multiple plots side-by-side (e.g., Training Loss vs. Validation Loss).

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 4))

ax1.plot(loss)
ax1.set_title("Loss")

ax2.plot(accuracy)
ax2.set_title("Accuracy")

6. The "Object-Oriented" vs "Pyplot" Debate​

  • plt.plot() (Pyplot): Great for quick interactive exploration.
  • fig, ax = plt.subplots() (OO style): Better for complex layouts and production scripts where you need fine-grained control over every element.

References for More Details​

Visualization is the final piece of our programming foundations. Now that you can process data with NumPy, clean it with Pandas, and visualize it with Matplotlib, you are ready to start building actual models.