🌳 Day 44: Dendrogram in Python

On Day 44 of our Data Visualization journey, we explored one of the most important visual tools in clustering  the Dendrogram.

If you’ve ever worked with hierarchical clustering or wanted to visually understand how data groups together, this chart is for you.

---

🎯 What is a Dendrogram?

A Dendrogram is a tree-like diagram used to visualize the results of Hierarchical Clustering.

It shows:

• How data points are grouped

• The order in which clusters merge

• The distance between clusters

• The hierarchical structure of data

Think of it as a family tree — but for data.

---

📊 What We’re Visualizing

In this example:

• We generate random data (10 data points, 4 features each)

• Apply hierarchical clustering

• Use the Ward linkage method

• Plot the cluster hierarchy as a dendrogram

---

🧑‍💻 Python Implementation

---

✅ Step 1: Import Libraries

import numpy as np
import matplotlib.pyplot as plt
from scipy.cluster.hierarchy import dendrogram, linkage

We use:

• NumPy → Generate sample dataset

• SciPy → Perform hierarchical clustering

• Matplotlib → Plot the dendrogram

---

✅ Step 2: Generate Sample Data

np.random.seed(42)
data = np.random.rand(10, 4)

• 10 observations

• 4 features per observation

• Random but reproducible

---

✅ Step 3: Apply Hierarchical Clustering

linked = linkage(data, method='ward')

Why Ward Method?

The Ward method minimizes variance within clusters.

It creates compact, well-separated clusters — ideal for structured grouping.

---

✅ Step 4: Plot the Dendrogram

plt.figure(figsize=(8, 5))
dendrogram(linked)
plt.title("Dendrogram - Hierarchical Clustering")
plt.xlabel("Data Points")
plt.ylabel("Distance") 
plt.show()

📈 Understanding the Output

In the dendrogram:

• Each leaf at the bottom represents a data point

• Vertical lines represent cluster merges

• The height of the merge shows distance between clusters

• The higher the merge, the less similar the clusters

Key Insight:

You can "cut" the dendrogram at a specific height to decide how many clusters you want.

For example:

• Cutting at a low height → many small clusters

• Cutting at a high height → fewer larger clusters

---

💡 Why Dendrograms Are Powerful

✔ Visualize cluster structure clearly
✔ Help decide optimal number of clusters
✔ Show similarity between data points
✔ Provide hierarchical relationships

---

🔥 Real-World Applications

• Customer segmentation

• Gene expression analysis

• Document clustering

• Product grouping

• Market research

• Image pattern recognition

---

🚀 When to Use a Dendrogram

Use it when:

• You want to understand data hierarchy

• The number of clusters is unknown

• You need explainable clustering

• You want visual validation of grouping