NETAJI SUBHASH ENGINEERING COLLEGE

NAME: SUBHAJIT NANDI

SEMESTER : 7th
TOPIC :
CLASS ROLL : 83
UNSUPERVISED MACHINE LEARNING
SECTION : B AND ITS APPLICATIONS
AND K-MEANS CLUSTERING
UNIVERSITY ROLL: 10900121089

STREAM: COMPUTER SCIENCE AND ENGINEERING

PAPER NAME : MACHINE LEARNING

PAPER CODE : PEC-CS701E

Introduction to
Unsupervised
Machine Learning
Unsupervised machine learning is a type of machine learning that allows computers
to learn without explicit labels or guidance. Instead of being fed labeled data,
unsupervised learning algorithms discover patterns and structures within the data
itself. This type of learning is advantageous when dealing with data that has not
been categorized or classified, enabling the identification of hidden patterns or
intrinsic structures.

Some common techniques used in unsupervised learning include clustering, association, and dimensionality reduction.
Clustering algorithms, such as K-means and hierarchical clustering, group data points based on similarity, helping to identify
natural groupings within the data. Association rule learning, like the Apriori algorithm, discovers interesting relationships
between variables in large databases. Dimensionality reduction techniques, such as Principal Component Analysis (PCA) and
t-distributed Stochastic Neighbor Embedding (t-SNE), reduce the number of random variables under consideration,
simplifying the dataset while preserving its essential features.
 Advantages of Unsupervised Learning
1 Discovery
Unsupervised learning excels at uncovering hidden patterns and insights that might
be overlooked in traditional methods. This ability to identify novel information is
invaluable for understanding complex datasets.

2 Efficiency
Unlike supervised learning, unsupervised learning doesn't
require human-labeled data, making it more efficient in
situations where labeled data is scarce or expensive to
obtain.
3 Flexibility
Unsupervised learning algorithms can adapt to different data
structures and uncover various patterns, making them
adaptable to a wide range of tasks.

4 Applications
Unsupervised learning has a wide range of applications in various fields, including
customer segmentation, fraud detection, and anomaly detection.
 Applications of Unsupervised Learning
Unsupervised learning can be used to identify distinct groups of customers based
Customer Segmentationon their purchasing behavior, demographics, or preferences, enabling businesses
to tailor marketing campaigns effectively.

In the realm of fraud detection, unsupervised learning algorithms can analyze transaction data to
spot irregularities that may indicate fraudulent activities. By clustering similar transactions together,
Anomaly Detection these algorithms can highlight outliers that deviate from normal behavior. Techniques such as
Isolation Forests, One-Class SVMs, and autoencoders are often used to detect these anomalies,
enabling financial institutions to proactively identify and prevent fraud.

Unsupervised learning algorithms play a crucial role in the field of image recognition by
clustering images based on their visual features. This capability enables various tasks, such as
Image Recognition image classification and object detection, without the need for extensive labeled datasets. By
automatically grouping similar images together, unsupervised learning can identify patterns
and structures that are not immediately apparent.
 Clustering Algorithms: K-Means Clustering
K-Means Clustering

K-Means is a popular and widely used

unsupervised learning algorithm for grouping
data points into clusters based on their
similarity. It aims to minimize the distance
between data points within the same cluster
and maximize the distance between clusters.

Centroid-Based
K-Means works by iteratively assigning data
points to the nearest cluster centroid and
recalculating the centroids based on the
assigned points, aiming for an optimal cluster
arrangement.
K-Means Clustering: Methodology and
Intuition
Initialization
Randomly select k initial centroids, representing the
center of each cluster.

Assignment
Assign each data point to the closest centroid based on a
distance metric, such as Euclidean distance.

Update
Recalculate the position of each centroid based on the
average of the data points assigned to that cluster.

Iteration
Repeat the assignment and update steps until the
centroids no longer change significantly, indicating that
the clusters have converged.
Advantages and Limitations of K-Means
Clustering
Advantages Limitations
If the initial centroids are not
The algorithm is straightforward to Sensitivity to well-chosen, the algorithm may
Simplicity and efficiencyimplement and understand, which converge to a local minimum,
makes it a popular choice for those initial centroid
resulting in suboptimal
new to machine learning. selection clustering.

It has been extensively studied In real-world scenarios, data may

Widely used and and documented, making it Assumption of not conform to this assumption,
well-understood in both spherical leading to poor clustering
well-understood
academic and industry settings. performance.
clusters
The algorithm is computationally efficient,
Suitable for large with a time complexity of O(n), where n is
Inability to handle Outliers can
datasets the number of data points. This efficiency
disproportionately affect the
allows K-Means to quickly process large noisy or outlier
position of centroids, leading
volumes of data, making it ideal for data effectively to skewed clustering results.
applications where real-time or near-real-
time analysis is required.
 Real-World Use Cases of K-Means
Clustering
Customer Segmentation
K-Means is widely used for customer segmentation, grouping
customers with similar characteristics to personalize marketing
efforts.

Image Compression
K-Means can be used to compress images by clustering similar colors, reducing the overall data size.

Document Clustering

Clustering similar documents based on their content helps with information retrieval and organization.

Medical Diagnosis
K-Means can be used to identify groups of patients with similar
symptoms and medical histories, aiding in diagnosis and treatment
planning.
 Conclusion and Future
Developments
Unsupervised learning has made significant strides in recent years, with
advancements in algorithms and techniques continually pushing the boundaries of
what is possible. The rapid growth of data in today's digital age has underscored the
importance of these methods, as they are uniquely capable of uncovering hidden
patterns and structures within vast, unlabelled datasets.

This ability to analyze and interpret data without the need for manual labeling has
made unsupervised learning methods increasingly valuable. They are now pivotal in
a wide range of applications, from market segmentation and customer behavior
analysis to anomaly detection and bioinformatics. As the volume and complexity of
data continue to expand, the role of unsupervised learning in driving innovation
and extracting meaningful insights becomes ever more critical.

Furthermore, the continuous evolution of unsupervised learning techniques, such

as advanced clustering methods, dimensionality reduction, and association rule
learning, has enhanced their robustness and applicability. These advancements
have enabled more accurate and efficient data analysis, empowering organizations
to make better-informed decisions and uncover opportunities that were previously
hidden