2. ROAD MAP:
• What is Machine Learning?
• What are the different types of Machine Learning?
• What is a Supervised Machine learning technique?
• What are the different types of Supervised learning
Problems?
• What are the algorithms used in each type of SLP? What
are their Strengths and Shortcomings?
3. What is Machine Learning?
• Study of computer
algorithms that improve
automatically through
experience.
• Learns and trains itself
using the training data.
• Tests itself using the
testing data to check the
accuracy of the prediction
or analysis.
4.
5. What are different types of Machine Learning?
• Supervised Learning
• Unsupervised Learning
• Reinforcement Learning
6. What is Supervised Machine Learning Technique?
• Trains from data that is
well labelled.
• Some data is already
tagged with the correct
answer.
• Allows collecting data and
produce data output from
the previous experiences.
7. What are the different types of Supervised learning
Problems?
1. Regression:
– Set of machine learning
methods that allow us to
predict a continuous
outcome variable.
– Aim of the model is to build
a mathematical equation
that defines y as a function
of the x variables.
8. What are the different types of Supervised learning
Problems?
2. Classification
– Predictive modeling
problem where a class label
is predicted for a given input
data.
– The output is prediction of
categorical values
9. What are the algorithms used in each type of SLP?
What are their Strengths and Shortcomings?
1. Regression:
• Linear Regression
• Decision Tree
• Support Vector Machines
2. Classification:
• Logistic Regression
• Decision Tree
• Naive Baiyes
10. LINEAR REGRESSION:
• One of the most common
algorithms used for
supervised regression
tasks.
• Linear Model that
assumes a linear
relationship between the
input variables and the
single output variable.
11. LINEAR REGRESSION:
• Strengths:
– Simple and Straightforward
– Can be regularized to avoid
overfitting
– Perfect for simple
regression problems.
– Proven to have high
accuracy when used for
linear relationships.
• Shortcomings:
– Performs poorly for non-
linear relationships.
– Not flexible enough to
capture more complex
problems.
12. DECISION TREE:
• Supervised Learning
algorithm that can be used
for both classification and
regression tasks.
• Continously split the
dataset into branches.
• Maximum information gain
from each branch.
13. DECISION TREE:
• Strengths:
– Good at learning non-linear
relationships.
– Able to handle both
continuous and categorical
variables.
– Perform classification
without requiring much
computation.
• Shortcomings
– Unconstrained decision
trees can be prone to
overfitting.
– Prone to errors in
classification problems with
many class and relatively
small number of training
examples.
14. Logistic Regression:
• Supervised learning
algorithm used for
Classification.
• Uses logistic function to
categorize data between 0
and 1.
• Uses a threshold value to
categorize.
15. Logistic Regression:
• Strengths
– Less inclined to over-fitting
– Good accuracy for many
simple data sets.
– Easier to implement,
interpret, and very efficient
to train.
• Shortcomings
– Tough to obtain complex
relationships with Logistic
Regression.
– Doesn't have good
accuracy in non-linear
relationships.
16. Naive Baiyes Algorithm:
• Simple Classification
Algorithm based on
Conditionial Probability.
• Based on a statistical
classification technique
called ‘Bayes Theorem’.
• Makes an assumption that
the predictor variables are
independent from each
other
17. Naive Baiyes Algorithm:
• Strengths
– Easy and quick way to
predict classes, both in
binary and multiclass
classification problems.
– Performs better compared
to other classification
models, even with less
training data.
• Shortcomings
– oversimplification leads to
possible outperforming.
– naive assumption of
independence is very
unlikely to match real-world
data.