Scribd
Upload
12 views10 pages

Beta

Uploaded by

Dibyangana Bose
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
12 views10 pages

Beta

Uploaded by

Dibyangana Bose
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

SUPERVISED MACHINE LEARNING AND ITS

APPLICATIONS AND DECISION TREE LEARNING

NAME: DIBYANGANA BOSE


STREAM: CSE , 4TH YEAR
SECTION: A
CLASS ROLL: 09
UNIVERSITY ROLL: 10900121009
SUBJECT : MACHINE LEARNING | PEC-CS701E
Introduction to Supervised
Machine Learning
1 Labeled Data
Supervised learning models are trained on datasets with labeled inputs
labeled inputs and expected outputs.

2 Predictive Models
The algorithms learn patterns in the data to make predictions on new,
unseen samples.

3 Classification and Regression


Supervised learning tasks can be divided into classification (predicting
categories) and regression (predicting numerical values).
Supervised Learning Algorithms
Linear Regression Logistic Regression Support Vector Decision Trees
Machines
Predicts a continuous Classifies data into Constructs a tree-like
numerical output based discrete categories, Finds the optimal model of decisions and
on one or more input such as true/false or hyperplane to separate their possible
features. Finds the best- spam/not spam. different classes of consequences. Can
fit line or hyperplane to Models the probability data. Works well with handle both
model the relationship of an output belonging high-dimensional data classification and
between the input and to a particular class. and can handle non- regression problems.
output variables. linear relationships.
Commonly used for Provides a clear,
Useful for forecasting, binary classification Effective for tasks like interpretable model for
price prediction, and problems like email image recognition, text making predictions and
other quantitative filtering or medical classification, and understanding the
tasks. diagnosis. bioinformatics. underlying data.
Regression and Classification
Tasks
Regression
Predicting continuous numerical values, such as house prices or stock prices.

Classification
Assigning data to discrete categories, such as email spam detection or image
recognition.

Hybrid Tasks
Some problems, like sentiment analysis, can involve both regression and
classification.
Decision Tree Learning
1 Recursive Partitioning
Decision trees work by recursively partitioning the data
based on the most informative features.

2 Greedy Algorithm
The algorithm greedily selects the feature that provides the
greatest information gain at each step.

3 Interpretable Models
Decision trees produce easily interpretable models, making
them useful for understanding the underlying patterns in
data.
Advantages of Decision Trees
1 Simplicity 2 Flexibility
Decision trees are intuitive and easy to They can handle both numerical and categorical
understand, even for non-technical stakeholders. features, and can model complex nonlinear
relationships.

3 Robustness 4 Scalability
Decision trees are relatively insensitive to outliers Decision tree algorithms can efficiently handle
and can handle missing data effectively. large-scale datasets and high-dimensional feature
spaces.
Building Decision Trees
Feature Selection
Choosing the most informative features to split the data
at each node.

Tree Growing
Recursively partitioning the data until a stopping criterion
is met.

Pruning
Removing unnecessary branches to improve
generalization and prevent overfitting.
Pruning Decision Trees

Cost Complexity
Pruning balances model complexity and training error to optimize performance.

Validation
Evaluating the pruned tree's performance on a held-out validation set.

Generalization
Pruning helps the model generalize better to new, unseen data.
Applications of Supervised Machine Learning
Image Recognition Classify images into different categories (e.g., object
detection, facial recognition).

Spam Detection Identify email messages as spam or legitimate based


on their content and metadata.

Predictive Maintenance Forecast when equipment might fail based on sensor


data and historical patterns.

Sentiment Analysis Determine the emotional sentiment (positive,


negative, neutral) of text data.
Conclusion and Future
Directions
Supervised machine learning has proven to be a transformative
technology with a wide range of real-world applications. As the field
continues to evolve, we can expect to see even more powerful and
versatile algorithms that can tackle increasingly complex problems.

You might also like