EXPERIMENT NO.

:-4

AIM: - Analyze and Develop Clustering Models for Dataset Exploration.

THEORY:-

Objective:
This experiment aims to provide with a thorough understanding of different clustering algorithms and
their practical applications.
Master the core concepts of K-Means, Hierarchical Clustering, and DBSCAN.
Implement these algorithms using Python and the scikit-learn library.
Conduct rigorous model evaluation using appropriate metrics.
Gain practical experience in data preprocessing, feature scaling, and model selection.
Analyze and interpret clustering results to draw meaningful insights.
Materials:
Software: Python with the following libraries:
pandas: For data manipulation and analysis.
numpy: For numerical computing.
scikit-learn: For implementing machine learning algorithms, including clustering.
matplotlib and seaborn: For data visualization.
IDE: Jupyter Notebook or any preferred Python development environment.
Datasets:
Iris Dataset: A classic, built-in dataset in scikit-learn, ideal for introductory clustering.
Customer Segmentation Dataset: A real-world dataset (e.g., from the UCI Machine
Learning Repository) to apply clustering in a practical scenario.
Procedure:
1. Data Loading and Preprocessing:
Load Data: Import the Iris and Customer Segmentation datasets into your chosen
environment using pandas.
Handle Missing Values:
Deletion: Remove rows or columns with missing values (if the proportion is
small).
Imputation: Replace missing values with estimated values (e.g., mean, median,
KNN imputation).
Feature Scaling: Standardize or normalize the data to ensure all features have
comparable scales:
Standardization (Z-score normalization): Transforms features to have zero
mean and unit variance.
Normalization (Min-Max scaling): Scales features to a specific range (e.g.,
between 0 and 1).
2. Implementation of Clustering Algorithms:
K-Means Clustering:
 Concept: Partitions data into 'k' clusters by minimizing the within-cluster sum of
squares (WCSS).
Implementation:
Use scikit-learn's KMeans class.
Experiment with different values of 'k' (number of clusters).
Elbow Method: Plot the WCSS against different 'k' values. The "elbow"
point (where the curve starts to bend) often indicates an optimal 'k'.
Silhouette Score: Calculate the Silhouette Score for each data point to
evaluate cluster cohesion and separation. Higher scores generally indicate
better-defined clusters.
Visualization: Create scatter plots to visualize the clusters in two-dimensional
space (if applicable). Color-code data points based on their assigned cluster.
Hierarchical Clustering:
Concept: Creates a hierarchical tree-like structure (dendrogram) representing the
relationships between data points.
Implementation:
Use scikit-learn's AgglomerativeClustering class.
Experiment with different linkage methods (single, complete, average) to
determine the most suitable approach for your data.
Dendrogram: Visualize the dendrogram to identify natural clusters by
cutting the tree at appropriate heights.
Visualization: Similar to K-Means, create scatter plots to visualize the clusters.
DBSCAN (Density-Based Spatial Clustering of Applications with Noise):
Concept: Groups together data points that are closely packed together (dense
regions) while ignoring outliers.
Implementation:
Use scikit-learn's DBSCAN class.
Tune the hyperparameters:
Epsilon (ε): The maximum distance between two samples for
them to be considered as in the same neighborhood.
Min_samples: The minimum number of samples in a
neighborhood for a point to be considered as a core point.
Visualization: Create scatter plots to visualize the clusters, highlighting core
points, border points, and noise points.
3. Model Evaluation:
Within-Cluster Sum of Squares (WCSS): For K-Means, calculate the WCSS for each
cluster. Lower WCSS generally indicates better clustering.
Silhouette Score: Calculate the Silhouette Score for all three algorithms to assess the
overall quality of clustering.
Visual Inspection: Carefully examine the scatter plots and dendrograms to assess the
interpretability and separation of clusters.
4. Comparison and Discussion:
 Tabulate the performance of each algorithm based on the evaluation metrics (WCSS,
Silhouette Score).
Discuss the advantages and disadvantages of each algorithm in terms of:
Assumptions: K-Means assumes spherical clusters, while DBSCAN can handle
irregularly shaped clusters.
Scalability: K-Means can be computationally expensive for large datasets.
Sensitivity to outliers: DBSCAN is more robust to outliers than K-Means.
Analyze the results and draw meaningful insights from the clustering. For example:
In customer segmentation, identify distinct customer groups based on their
purchasing behavior.
In biological data analysis, discover natural groupings of species or cell types.

Example Output (K-Means Clustering on Iris Dataset):

Python
from sklearn.cluster import KMeans
from sklearn.datasets import load_iris
import matplotlib.pyplot as plt

iris = load_iris()
X = iris.data[:, :2] # Use only the first two features for visualization

# Determine optimal number of clusters using the Elbow Method

wcss = []
for i in range(1, 11):
kmeans = KMeans(n_clusters=i, init='k-means++', random_state=42)
kmeans.fit(X)
wcss.append(kmeans.inertia_)

plt.plot(range(1, 11), wcss)

plt.title('Elbow Method for Iris Dataset')
plt.xlabel('Number of clusters (k)')
plt.ylabel('Within-Cluster Sum of Squares (WCSS)')
plt.show()

# Apply K-Means with the optimal number of clusters (e.g., k=3)

kmeans = KMeans(n_clusters=3, init='k-means++', random_state=42)
y_kmeans = kmeans.fit_predict(X)

# Visualize the clusters

plt.scatter(X[y_kmeans == 0, 0], X[y_kmeans == 0, 1], s=100, c='red',
label='Cluster 1')
plt.scatter(X[y_kmeans == 1, 0], X[y_kmeans == 1, 1], s=100, c='blue',
label='Cluster 2')
plt.scatter(X[y_kmeans == 2, 0], X[y_kmeans == 2, 1], s=100, c='green',
label='Cluster 3')
plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1], s=300,
c='yellow', label='Centroids')
plt.title('K-Means Clustering on Iris Dataset')
plt.xlabel('Sepal Length')
plt.ylabel('Sepal Width')
plt.legend()
plt.show()

Extensions:
Explore other clustering algorithms like Gaussian Mixture Model (GMM).
Implement more advanced data preprocessing techniques, such as dimensionality reduction
(PCA).
Conduct a more in-depth analysis and visualization of the results, including 3D scatter plots
and interactive visualizations.
Apply clustering to different domains, such as image segmentation, anomaly detection, and
social network analysis.

RESULT: - Thus, all the various clustering models on dataset exploration have been executed
successfully.

PRACTICAL ASSIGNMENT:
1. Explain the concept of hierarchical clustering. How does it differ from partitional clustering
methods like K-Means?
2. Describe the K-Means clustering algorithm. What are its key steps and how does it determine the
optimal number of clusters?