Chapter 5: Unsupervised Learning

🧠 In Unsupervised Learning, the model discovers patterns or structures from data without any labels.
No answers are given — the algorithm finds hidden relationships on its own.

---

🔍 Key Idea:

“Group similar things together or reduce complexity without knowing the correct output.”

---

✳️ Types of Unsupervised Learning

Type	Description	Use Case
Clustering	Group similar data points	Customer Segmentation
Dimensionality Reduction	Reduce features while preserving information	Data Visualization
Association Rule Learning	Find rules between variables	Market Basket Analysis

---

🔹 1. Clustering

Grouping data points such that those in the same group are more similar to each other than to those in other groups.

a. K-Means Clustering

• Partitions data into K clusters

• Iteratively updates cluster centers (centroids)

python
from sklearn.cluster import KMeans
model = KMeans(n_clusters=3)
model.fit(X)
print(model.labels_)

📌 Example: Grouping customers by spending habits

---

b. Hierarchical Clustering

• Builds a tree (dendrogram) of clusters

• Doesn’t require pre-specifying number of clusters

📌 Example: Document or gene clustering

---

🔹 2. Dimensionality Reduction

Reducing the number of features (columns) in the dataset while retaining important info

a. PCA (Principal Component Analysis)

• Converts high-dimensional data into fewer dimensions (components)

python
from sklearn.decomposition import PCA
pca = PCA(n_components=2)
X_reduced = pca.fit_transform(X)

📌 Used for: Visualization, noise removal, speeding up algorithms

---

b. t-SNE

• Better for visualizing high-dimensional data

• Preserves local structure (similar points stay close)

📌 Often used to visualize clusters in NLP or image features

---

🔹 3. Association Rule Learning

Finds relationships between variables in large datasets.

a. Apriori Algorithm

• Discovers frequent itemsets and rules

📌 Example: If a customer buys bread & butter, they’re likely to buy milk.

b. Eclat Algorithm

• More memory-efficient than Apriori

• Uses set intersections

📌 Used in: Market Basket Analysis, Recommender Systems

---

📏 Evaluation Metrics in Unsupervised Learning

Since there's no "correct label", metrics are indirect.

Metric	Use
Silhouette Score	How well-clustered data points are
Inertia (for KMeans)	Measures compactness of clusters
Explained Variance (PCA)	How much info each component keeps

---

💻 Hands-On Projects

✅ Customer Segmentation (K-Means)

• Data: Age, Spending Score, Income

• Output: Group customers into clusters

✅ Market Basket Analysis (Apriori)

• Input: Transaction data

• Output: Discover item combinations often bought together

---

🧠 Summary of Chapter 5

Concept	Summary
Unsupervised Learning	No labels; model finds patterns
Clustering	K-Means, Hierarchical
Dimensionality Reduction	PCA, t-SNE
Association Rules	Apriori, Eclat
Hands-on	Customer groups, Market rules

---

✅ Mini Assignment:

1. Use K-Means on a dataset with 2 features and visualize the clusters.

2. Apply PCA to reduce a dataset to 2 dimensions.

3. Try using the Apriori algorithm on transaction data.