ML U1 & U2 Notes
ML U1 & U2 Notes
Uploaded by
rehmahahmed6Copyright:
Available Formats
ML U1 & U2 Notes
ML U1 & U2 Notes
Uploaded by
rehmahahmed6Original Title
Copyright
Available Formats
Share this document
Did you find this document useful?
Is this content inappropriate?
Copyright:
Available Formats
ML U1 & U2 Notes
ML U1 & U2 Notes
Uploaded by
rehmahahmed6Copyright:
Available Formats
Machine Learning Unit 1 and Unit 2 Notes
Unit 1
Q1) What do you mean by Learning? Justify cognitive automation is
a subset of machine learning by giving examples.
es
● Machinelearningis a process through which computerized
systems use human-supplied data and feedback to independently
ot
make decisions and predictions, typically becoming more accurate
with continual training. This contrasts with traditional computing,
in which every action taken by a computer must be
pre-programmed.
N
● Reinforcementlearningteaches a system as it interacts with an
's
environment by offering it rewards when it performs an action
correctly.
ah
● Supervisedlearning, which applies to the computer-vision systems
used in autonomous vehicles
● Unsupervisedlearning, which is used when data need to be
hm
clustered (for example, audience segmentation for streaming
services or product recommendations to online shoppers).
● ’Cognition’refers to all processes by which the sensory input is
Re
transformed, reduced, elaborated, stored, recovered, and used.
Such terms as sensation, perception, imagery, retention, recall,
problem-solving, and thinking, among many others, refer to
hypothetical stages or aspects of cognition
● Automationrefers to the full or partial “execution by a machine
agent (usually a computer) of a function that was previously
carried out by a human”
● Cognitive Automationrefers to seizing ML for automating
cognitive knowledge and service work to realize value offered by
AI, which is based on implementing artificial cognition that
mimics and approximates human cognition in machines
Justifying Cognitive Automation as a Subset of Machine Learning with
Examples:
es
● Natural Language Processing (NLP) in Customer Service: Cognitive
automation systems that handle customer inquiries use supervised
ot
machine learning models to understand text (via NLP). They learn
from historical chat data to provide responses that mimic a
human customer service agent. These systems can independently
N
refine their responses based on interactions, making them more
efficient over time.
's
● Example: Virtual assistants like chatbots that interpret customer
queries and provide responses, learning from past interactions to
ah
become better at understanding and solving customer problems.
● Automated Document Processing in Finance: Cognitive automation
can read and understand invoices, contracts, and financial
hm
statements using machine learning. It uses techniques from
supervised learning to extract information from scanned
documents and reinforcement learning to improve over time based
Re
on the accuracy of its extractions.
● Example: A machine learning system automating the review and
approval of loan applications, making decisions by learning from
historical data on past approvals and rejections.
● Intelligent Decision Support Systems: Machine learning
algorithms in these systems learn from historical data and human
input to assist in decision-making. For instance, in healthcare,
cognitive automation helps doctors by recommending treatment
plans based on previous patient data, imaging, and test results,
which uses both unsupervised learning (to group similar patient
cases) and supervised learning (to make treatment predictions).
● Example: AI-powered diagnostic tools that assist doctors by
analyzing medical data and images to suggest possible diagnoses.
es
Q2) Describe various phases used by machine learning.
ot
N
's
ah
hm
Re
1. Gathering Data:
Data Gathering is the first step of the machine learning life cycle. The
goal of this step is to identify and obtain all data-related problems.
This step includes the below tasks:
● Identify various data sources
● Collect data
● Integrate the data obtained from different sources
● By performing the above task, we get a coherent set of data, also
called as a dataset. It will be used in further steps.
2. Data preparation
After collecting the data, we need to prepare it for further steps.
Data preparation is a step where we put our data into a suitable place
es
and prepare it to use in our machine learning training.
This step can be further divided into two processes:
ot
Data exploration:
It is used to understand the nature of data that we have to work with.
We need to understand the characteristics, format, and quality of
N
data. A better understanding of data leads to an effective outcome. In
this, we find Correlations, general trends, and outliers.
's
Data pre-processing:
Now the next step is preprocessing of data for its analysis.
ah
3. Data Wrangling
Data wrangling is the process of cleaning and converting raw data into a
hm
useable format. It is the process of cleaning the data, selecting the
variable to use, and transforming the data in a proper format to make
it more suitable for analysis in the next step.
Re
● Missing Values
● Duplicate data
● Invalid data
● Noise
4. Data Analysis
Now the cleaned and prepared data is passed on to the analysis step.
This step involves:
● Selection of analytical techniques
● Building models
● Review the result
The aim of this step is to build a machine learning model to analyze the
data using various analytical techniques and review the outcome.
es
Hence, in this step, we take the data and use machine learning
algorithms to build the model.
ot
5. Train Model
Now the next step is to train the model, in this step we train our model
N
to improve its performance for better outcome of the problem.
We use datasets to train the model using various machine learning
's
algorithms. Training a model is required so that it can understand the
various patterns, rules, and, features.
ah
6. Test Model
Once our machine learning model has been trained on a given dataset,
hm
then we test the model. In this step, we check for the accuracy of our
model by providing a test dataset to it.
Testing the model determines the percentage accuracy of the model as
Re
per the requirement of project or problem.
7. Deployment
The last step of machine learning life cycle is deployment, where we
deploy the model in the real-world system.
If the above-prepared model is producing an accurate result as per our
requirement with acceptable speed, then we deploy the model in the
real system. But before deploying the project, we will check whether it
is improving its performance using available data or not. The
deployment phase is similar to making the final report for a project.
Q3) Every machine learning algorithm should have some key points
while designing. What are they? Explain them in brief.
es
ot
N
's
ah
hm
Step 1) Choosing the Training Experience:The very important and
Re
first task is to choose the training data or training experience which
will be fed to the Machine Learning Algorithm. It is important to note
that the data or experience that we fed to the algorithm must have a
significant impact on the Success or Failure of the Model. So Training
data or experience should be chosen wisely.
Below are the attributes which will impact on Success and Failure of
Data:
● The training experience will be able to provide direct or indirect
feedback regarding choices. For example: While Playing chess the
training data will provide feedback to itself like instead of this
move if this is chosen the chances of success increases.
● Second important attribute is the degree to which the learner
es
will control the sequences of training examples. For example:
when training data is fed to the machine then at that time
ot
accuracy is very less but when it gains experience while playing
again and again with itself or opponent the machine algorithm will
get feedback and control the chess game accordingly.
N
● Third important attribute is how it will represent the distribution
of examples over which performance will be measured. For
's
example, a Machine learning algorithm will get experience while
going through a number of different cases and different
ah
examples. Thus, Machine Learning Algorithm will get more and
more experience by passing through more and more examples and
hence its performance will increase.
hm
Step 2- Choosing target function:The next important step is
choosing the target function. It means according to the knowledge fed
Re
to the algorithm the machine learning will choose NextMove function
which will describe what type of legal moves should be taken. For
example : While playing chess with the opponent, when opponent will
play then the machine learning algorithm will decide what be the
number of possible legal moves taken in order to get success.
Step 3- Choosing Representation for Target function:When the
machine algorithm will know all the possible legal moves the next step is
to choose the optimized move using any representation i.e. using linear
Equations, Hierarchical Graph Representation, Tabular form etc. The
NextMove function will move the Target move like out of these move
which will provide more success rate. For Example : while playing chess
machine have 4 possible moves, so the machine will choose that
es
optimized move which will provide success to it.
ot
Step 4- Choosing Function Approximation Algorithm:An optimized
move cannot be chosen just with the training data. The training data
had to go through with set of example and through these examples the
N
training data will approximates which steps are chosen and after that
machine will provide feedback on it. For Example : When a training data
's
of Playing chess is fed to algorithm so at that time it is not machine
algorithm will fail or get success and again from that failure or success
ah
it will measure while next move what step should be chosen and what is
its success rate.
hm
Step 5- Final Design:The final design is created at last when system
goes from number of examples , failures and success , correct and
incorrect decision and what will be the next step etc. Example:
Re
DeepBlue is an intelligent computer which is ML-based won chess game
against the chess expert Garry Kasparov, and it became the first
computer which had beaten a human chess expert.
Q4) What is the process of machine learning algorithm and its
testing in real life?
Q5) State and explain various types of data used in ML with
suitable examples.
Types of Data related to Machine Learning
Data Types are a way of classification that specifies which type of
value a variable can store and what type of mathematical operations,
relational, or logical operations can be applied to the variable without
es
causing an error. In Machine learning, it is very important to know
appropriate datatypes of independent and dependent variable.
ot
as it provides the basis for selecting classification or regression
models. Incorrect identification of data types leads to incorrect
modeling which in turn leads to an incorrect solution.
N
Here I will be discussing different types of data types with suitable
examples.
's
Different Types of data types
ah
hm
Re
1. Quantitative data type: –
This type of data type consists of numerical values. Anything which is
measured by numbers.
E.g., Profit, quantity sold, height, weight, temperature, etc.
This is again of two types
A.) Discrete data type: –
The numeric data which have discrete values or whole numbers. This
type of variable value if expressed in decimal format will have no
proper meaning. Their values can be counted.
es
E.g.: – No. of cars you have, no. of marbles in containers, students in a
class, etc.
ot
N
's
ah
B.) Continuous data type: –
The numerical measures which can take the value within a certain
range. This type of variable value if expressed in decimal format has
hm
true meaning. Their values can not be counted but measured. The value
can be infinite
E.g.: – height, weight, time, area, distance, measurement of rainfall,
Re
etc.
es
2. Qualitative data type: –
ot
These are the data types that cannot be expressed in numbers. This
describes categories or groups and is hence known as the categorical
data type.
This can be divided into:-
N
's
a. Structured Data:
This type of data is either number or words. This can take numerical
ah
values but mathematical operations cannot be performed on it. This
type of data is expressed in tabular format.
hm
E.g.) Sunny=1, cloudy=2, windy=3 or binary form data like 0 or1, Good
or bad, etc.
Re
b. Unstructured data:
This type of data does not have the proper format and therefore
known as unstructured data.This comprises textual data, sounds,
images, videos, etc.
es
ot
N
Besides this, there are also other types refer as Data Types
preliminaries or Data Measures:-
's
These can also be refer different scales of measurements.
ah
I. Nominal Data Type:
This is in use to express names or labels which are not order or
hm
measurable.
E.g., male or female (gender), race, country, etc.
Re
II. Ordinal Data Type:
This is also a categorical data type like nominal data but has some
natural ordering associated with it.
E.g., Likert rating scale, Shirt sizes, Ranks, Grades, etc.
es
ot
III. Interval Data Type:
This is numeric data which has proper order and the exact zero means
N
the true absence of a value attached. Here zero means not a complete
absence but has some value. This is the local scale.
's
E.g., Temperature measured in degree Celsius, time, Sat score, credit
score, pH, etc. difference between values is familiar. In this case,
ah
there is no absolute zero. Absolute
hm
Re
IV. Ratio Data Type:
This quantitative data type is the same as the interval data type but
has the absolute zero. Here zero means complete absence and the
scale starts from zero. This is the global scale.
E.g., Temperature in Kelvin, height, weight, etc.
es
ot
N
Q6) Define the term dataset. What are the properties of dataset one
's
should consider while choosing dataset.
ah
A Dataset is a set of data grouped into a collection with which
developers can work to meet their goals. In a dataset, the rows
hm
represent the number of data points and the columns represent the
features of the Dataset. Datasets may vary in size and complexity and
they mostly require cleaning and preprocessing to ensure data quality
Re
and suitability for analysis or modeling.
Let us see an example below:
es
Dataset
This is the Iris dataset. Since this is a dataset with which we build
ot
models, there are input features and output features. Here:
The input features are Sepal Length, Sepal Width, Petal Length, and
Petal Width.
Species is the output feature.
N
's
Datasets can be stored in multiple formats. The most common ones are
CSV, Excel, JSON, and zip files for large datasets such as image
ah
datasets.
hm
Why are datasets used?
Datasets are used to train and test AI models, analyze trends, and gain
insights from data. They provide the raw material for computers to
Re
learn patterns and make predictions.
Types of Datasets
Numerical Dataset, Categorical Dataset, Web Dataset, Time series
Dataset, Image Dataset, Ordered Dataset, Partitioned Dataset,
File-Based Datasets, Bivariate Dataset, Multivariate Dataset
Data Interpretation
It means conducting a complete study of the data. Analyzing number of
rows and columns, data types, useful and redundant data, checking for
null values.
Based on this study various operations are done on the data to make it
suitable for entering in ML models. Operations such as Feature
Engineering, Dimension Reduction, Null Values induction, Missing values
es
fill-in, data types conversion by encoding methods, etc.
ot
The choice of dataset can significantly impact the model's
performance, generalization, and the insights it can provide. Here are
some key considerations and steps to guide you in choosing the right
N
dataset for your machine learning project:
's
1. Define Your Problem and Objectives:
Start by clearly defining the problem you want to solve and the
ah
objectives you want to achieve with your machine learning model.
Understanding the problem domain and the goals of your project is
essential for selecting an appropriate dataset.
hm
2. Data Relevance:
Ensure that the dataset is relevant to your problem. It should contain
Re
features (attributes) that are meaningful and related to the problem
you're trying to solve. Irrelevant or redundant features can introduce
noise and reduce model performance.
3. Data Size:
Consider the size of the dataset. In general, larger datasets tend to
produce more accurate and robust models, especially for complex
problems. However, collecting and processing large datasets can be
resource-intensive.
4. Data Quality:
Data quality is paramount. Check for missing values, outliers, and errors
in the dataset. Low-quality data can lead to biased or inaccurate
models. Data preprocessing may be required to clean and prepare the
es
dataset.
ot
5. Data Balance:
For classification problems, check the class distribution. An imbalanced
dataset (where one class significantly outnumbers the others) can lead
N
to biased models. Techniques like oversampling or undersampling may
be needed to address class imbalance.
's
6. Data Diversity:
ah
Ensure that the dataset covers a diverse range of scenarios or
conditions relevant to your problem. Diversity helps the model
generalize better to unseen data.
hm
7. Data Availability:
Consider the accessibility and availability of the dataset. Ensure that
Re
you have the necessary permissions to use the data, and check for any
legal or ethical constraints.
8. Data Collection:
Depending on your problem, you may need to collect your own data
through surveys, sensors, web scraping, or other means. Be mindful of
data collection methods and ethics.
9. Public Datasets:
Explore publicly available datasets from sources like Kaggle, UCI
Machine Learning Repository, government databases, or academic
datasets. These datasets can be a valuable resource for
experimentation.
es
10. Domain Knowledge:
- Leverage domain knowledge and expertise in the field related to your
ot
problem. Subject matter experts can guide you in selecting relevant
datasets and understanding the nuances of the data.
11. Data Exploration:
N
- Perform exploratory data analysis (EDA) to gain insights into the
's
dataset. Visualizations, summary statistics, and correlations can help
you understand the data's characteristics.
ah
12. Data Splitting:
- Divide the dataset into training, validation, and testing sets. This is
hm
crucial for model evaluation and preventing overfitting.
13. Ethical Considerations:
Re
- Be aware of ethical considerations when working with data, especially
if the data contains sensitive information. Ensure that privacy and
ethical guidelines are followed.
14. Data Licensing:
- Check the licensing terms and restrictions associated with the
dataset. Some datasets may have specific usage terms that you need
to adhere to.
15. Iterative Process:
- Dataset selection is often an iterative process. You may need to
experiment with different datasets to find the one that works best
es
for your problem.
ot
Q7) Compare supervised and unsupervised learning
N
Feature Supervised Learning Unsupervised Learning
Definition Learns from labeled data, Learns from unlabeled
where the correct output data, discovering
's
is known. patterns on its own.
ah
Objective Predict outcomes or Find hidden patterns,
classify data into group data, or detect
predefined categories. anomalies.
hm
Training Data Uses labeled data (data Uses unlabeled data
tagged with the correct (data without predefined
Re
output). labels).
Types of - Classification - Clustering (grouping
Problems (categorical outcomes) based on similarity)
- Regression (continuous - Association (finding
outcomes) relationships)
Examples - Predicting if an email is - Grouping customers by
spam or not purchasing behavior
- Predicting house prices - Market basket analysis
Algorithms - Linear/Logistic - K-Means Clustering
Regression - Hierarchical Clustering
- Decision Trees - Apriori Algorithm
es
- Support Vector - DBSCAN
Machines (SVM)
- Random Forests
ot
- Naive Bayes
N
Evaluation - Accuracy - Silhouette Score
Metrics - Precision - Adjusted Rand Index
- Recall - Davies-Bouldin Index
's
- F1 Score - Calinski-Harabasz
- Mean Squared Error Score
ah
(MSE)
Presence of Yes, output labels are No, output labels are not
hm
Output Labels present. available.
Task Maps input to a known Identifies hidden
output. patterns or structures in
Re
data.
Approach Learns by example with Learns without guidance,
guidance. based on data's inherent
structure.
Use Cases - Spam detection - Anomaly detection
- Fraud detection - Market segmentation
- Speech recognition - Network analysis
Advantages - Accurate predictions - Works with unlabeled
with well-labeled data data
- Can handle complex - Finds hidden patterns
es
tasks automatically
Disadvantages - Requires labeled data - May produce less
ot
- Time-consuming to label accurate results
data - Harder to interpret
N
- Can struggle with and validate findings
complex or dynamic
environments
's
Q8) Justify the need of data mining in machine learning.
ah
Data miningandmachine learningare closely interrelated,and data
mining is crucial for the success of machine learning algorithms. Here's
hm
a justification for the need of data mining in machine learning:
1. Extracting Relevant Patterns and Knowledge from Large Datasets
Re
● Machine learningalgorithms rely heavily on clean,
well-structured, and relevant data to learn and make predictions.
Data mininghelps in identifying and extracting usefulpatterns,
relationships, and trends from vast amounts of raw data, turning
it into meaningful information. For example, it can identify
patterns in customer behavior, fraud detection, or market
trends, which are valuable inputs for machine learning models.
2. Data Preprocessing
● Before machine learning algorithms can be applied, the raw data
must be preprocessed to handle missing values, outliers, and
irrelevant features.Data mining techniqueslike datacleaning,
es
transformation, and normalization ensure the dataset is of high
quality and suitable for training machine learning models.
ot
● For example, in customer data, irrelevant features such as
unrelated columns or inconsistencies in user details can negatively
affect model accuracy. Data mining helps clean and prepare such
data.
N
3. Feature Selection and Engineering
's
● The performance of machine learning algorithms significantly
ah
depends on the choice of features.Data miningtechniquesassist
in feature selection (choosing the most relevant variables) and
feature engineering (creating new useful features from existing
hm
ones). This improves the performance and accuracy of machine
learning models.
● Example: In predictive analytics for loan approvals, data mining
Re
can help identify key features like income, credit score, and loan
history, which are most relevant for making predictions.
4. Handling Unstructured Data
● A large portion of data, such as text, images, and videos, is
unstructured.Data miningtechniques can be used toextract
structured information from this data, which is essential for
feeding into machine learning models.
● Example: Mining text data to extract useful features such as
sentiment or topic, which can be used in natural language
processing (NLP) tasks like sentiment analysis or recommendation
systems.
es
5. Discovering Hidden Patterns in Unlabeled Data
● Unsupervised learningmethods (such as clustering or association)
ot
are used in machine learning to find hidden patterns in data
without labels.Data miningtechniques help in uncoveringthese
patterns, associations, or clusters, which machine learning models
N
can use to improve their decision-making processes.
● Example: In market segmentation, data mining might discover
's
clusters of customers with similar buying habits, which can then
be used for personalized marketing strategies.
ah
6. Reducing Dimensionality
● Data miningcan help in dimensionality reduction, which reduces
hm
the number of variables while preserving the important
information. This is essential for machine learning, especially
when dealing with high-dimensional data, where too many features
Re
can lead to overfitting or slow processing.
● Example: Techniques likePrincipal Component Analysis(PCA)or
Singular Value Decomposition (SVD)are often usedin data
mining to reduce dimensions, making machine learning algorithms
more efficient.
7. Improving Accuracy and Predictive Power
● By extracting useful patterns, relationships, and trends from raw
data,data miningenhances the accuracy and predictivepower of
machine learning models. Data mining helps in identifying the most
relevant factors that can influence outcomes, thereby refining
the learning process.
● Example: In fraud detection, data mining can uncover subtle
patterns in transaction data that machine learning algorithms can
es
use to accurately predict and detect fraud.
8. Knowledge Discovery in Databases (KDD)
ot
● Machine learning is often a part of the broader process called
Knowledge Discovery in Databases (KDD), where datamining
N
plays a pivotal role. Data mining techniques are essential to
discover previously unknown relationships or patterns from large
's
datasets, which can then be leveraged by machine learning models
for prediction and classification.
ah
● Example: In medical research, data mining can help discover
hidden correlations between symptoms and diseases, leading to
the development of better predictive healthcare models using
hm
machine learning.
9. Handling Noisy and Incomplete Data
Re
● Real-world data is often noisy or incomplete, which can degrade
the performance of machine learning models.Data mining
techniques are designed to clean and handle such data, ensuring
that the final dataset used for machine learning is reliable and
robust.
● Example: In financial data, missing values or errors in
transactional records can be corrected or imputed through data
mining techniques before applying machine learning for credit risk
assessment.
10. Efficiency in Dealing with Large-Scale Data
● With the growing amount of data generated every day (e.g., from
es
IoT devices, social media, or transactions),dataminingis
essential for handling and processing large-scale datasets
ot
efficiently. Machine learning models often need summarized or
reduced data to operate effectively, which data mining provides
through aggregation and summarization techniques.
N
● Example: In social media analysis, mining relevant social posts
from millions of records and summarizing them for machine
's
learning sentiment analysis or trend prediction.
Conclusion:
ah
Data miningserves as a crucial preprocessing step and knowledge
discovery tool in the machine learning pipeline. It ensures that the raw
hm
data is converted into a form that machine learning algorithms can
effectively learn from, enhancing their predictive accuracy, efficiency,
and relevance. Without data mining, machine learning would struggle to
Re
handle large-scale, unstructured, and noisy datasets, limiting its
real-world applicability.
Q9) As a researcher you are expected to work with detection of
COVID at early stage. What kind of data and types of data you
will select? What kind of properties you will consider while choosing
data and dataset.
Main detection features will be the following 3:
Antibody test (IgG)
Antibody testing is also known as serological testing. Your doctor or
medical laboratory technician will use it to examine the type of
antibodies present in your blood.
es
There are numerous antibodies in the blood. The technician or nurse
will collect a sample of your blood and examine it for IgM and IgG. Ig
ot
stands for an immunoglobulin molecule.
● IgM antibodies develop at an early stage of infection against
SARS-CoV-2.
N
● IgG antibodies develop against SARS-CoV-2 once the person has
recovered from coronavirus.
's
Results by Antibody Test (IgG)
ah
The antibody testing kits take around 30-60 minutes to show results.
Reverse Transcription Polymerase Chain Reaction (RT – PCR)
hm
A polymerase chain reaction test is a highly sensitive test. Due to its
increased sensitivity and high fidelity, it is known as the most accurate
testing method for COVID 19 to date. It works by detecting the
Re
presence of genetic material from a specific pathogen.
Results by RT-PCR
RT-PCR is capable of delivering an accurate diagnosis and result for
COVID 19 within 3 hours. The laboratories take 6-8 hours to derive a
conclusive result.
TrueNat
TrueNat is a chip-based, portable RT-PCR machine that was initially
developed to diagnose tuberculosis. You can confirm your sample using
confirmatory tests for SARS-CoV-2 if you test positive by TrueNat
Beta CoV.
Results by TrueNat
es
It is capable of producing faster results than standard RT-PCR tests.
ot
Aside from these, patient demographic and time for test and reporting
will also be recorded and used for detecting underlying patterns.
These underlying patterns can be used in statistics to generate
hypotheses and theories.
N
's
Properties of the data will be the same as Q6.
ah
Q10) What is the need of regression? Describe various types of
the same.
hm
● Regression Analysis is a statistical process for estimating the
relationships between the dependent variables or criterion
variables and one or more independent variables or predictors.
Re
● Regression analysis is generally used when we deal with a dataset
that has the target variable in the form of continuous data.
Regression analysis explains the changes in criteria about changes
in select predictors.
● The conditional expectation of the criteria is based on predictors
where the average value of the dependent variables is given when
the independent variables are changed.
● Three major uses for regression analysis are determining the
strength of predictors, forecasting an effect, and trend
forecasting.
● There are times when we would like to analyze the effect of
different independent features on the target or what we say
dependent features. This helps us make decisions that can affect
the target variable in the desired direction.
es
● Regression analysis is heavily based on statistics and hence gives
quite reliable results to this reason only regression models are
ot
used to find the linear as well as non-linear relation between the
independent and the dependent or target variables.
N
Types of Regression are as follows:
● Linear regressionis used for predictive analysis. Linear
's
regression is a linear approach for modeling the relationship
between the criterion or the scalar response and the multiple
ah
predictors or explanatory variables. Linear regression focuses on
the conditional probability distribution of the response given the
values of the predictors. The formula for linear regression is: y =
hm
θx + b
● Polynomial Regression:This is an extension of linear regression
and is used to model a non-linear relationship between the
Re
dependent variable and independent variables. Here as well
syntax remains the same but now in the input variables we include
some polynomial or higher degree terms of some already existing
features as well. Linear regression was only able to fit a linear
model to the data at hand but with polynomial features, we can
easily fit some non-linear relationship between the target as well
as input features.
● Stepwise regressionis used for fitting regression models with
predictive models. It is carried out automatically. With each step,
the variable is added or subtracted from the set of explanatory
variables. The approaches for stepwise regression are forward
selection, backward elimination, and bidirectional elimination. The
formula for stepwise regression is:
● Decision Tree Regression:A Decision Tree is the most powerful
es
and popular tool for classification and prediction. A Decision tree
is a flowchart-like tree structure, where each internal node
ot
denotes a test on an attribute, each branch represents an
outcome of the test, and each leaf node (terminal node) holds a
class label. There is a non-parametric method used to model a
N
decision tree to predict a continuous outcome.
● Random Forestis an ensemble technique capable of performing
's
both regression and classification tasks with the use of multiple
decision trees and a technique called Bootstrap and Aggregation,
ah
commonly known as bagging. The basic idea behind this is to
combine multiple decision trees in determining the final output
rather than relying on individual decision trees.
hm
● Support vector regression (SVR)is a type of support vector
machine (SVM) that is used for regression tasks. It tries to find
a function that best predicts the continuous output value for a
Re
given input value.SVR can use both linear and non-linear kernels. A
linear kernel is a simple dot product between two input vectors,
while a non-linear kernel is a more complex function that can
capture more intricate patterns in the data. The choice of kernel
depends on the data’s characteristics and the task’s complexity.
● Ridge Regression: Ridge regression is a technique for analyzing
multiple regression data. When multicollinearity occurs, least
squares estimates are unbiased. This is a regularized linear
regression model, it tries to reduce the model complexity by
adding a penalty term to the cost function. A degree of bias is
added to the regression estimates, and as a result, ridge
regression reduces the standard errors.
es
● Lasso regressionis a regression analysis method that performs
ot
both variable selection and regularization. Lasso regression uses
soft thresholding. Lasso regression selects only a subset of the
N
provided covariates for use in the final model. This is another
regularized linear regression model, it works by adding a penalty
term to the cost function, but it tends to zero out some features’
's
coefficients, which makes it useful for feature selection.
● ElasticNet Regression: Linear Regression suffers from
ah
overfitting and can’t deal with collinear data. When there are
many features in the dataset and even some of them are not
hm
relevant to the predictive model. This makes the model more
complex with a too-inaccurate prediction on the test set (or
overfitting). Such a model with high variance does not generalize
Re
on the new data. So, to deal with these issues, we include both
L-2 and L-1 norm regularization to get the benefits of both Ridge
and Lasso at the same time. The resultant model has better
predictive power than Lasso
● Bayesian Linear Regression: As the name suggests this algorithm
is purely based on Bayes Theorem. Because of this reason only we
do not use the Least Square method to determine the
coefficients of the regression model. So, the technique which is
used here to find the model weights and parameters relies on
features posterior distribution and this provides an extra
stability factor to the regression model which is based on this
technique.
Q11) Problems based on regression
es
Numerical PDF
ot
Q12) Note on polynomial regression
Polynomial Regression
N
Polynomial Regression is a regression algorithm that models the
relationship between a dependent(y) and independent variable(x) as nth
's
degree polynomial. The Polynomial Regression equation is given below:
ah
2 3 𝑛
𝑦 = 𝑏𝑜 + 𝑏
1𝑥 + 𝑏1𝑥 + 𝑏2𝑥
+ ... + 𝑏𝑛𝑥
It is also called the special case of Multiple Linear Regression in ML.
hm
Because we add some polynomial terms to the Multiple Linear
regression equation to convert it into Polynomial Regression.
It is a linear model with some modification in order to increase the
Re
accuracy. The dataset used in Polynomial regression for training is of
non-linear nature. It makes use of a linear regression model to fit the
complicated and nonlinear functions and datasets.
Hence, "In Polynomial regression, the original features are converted
into Polynomial features of required degree (2,3,..,n) and then modeled
using a linear model."
Need for Polynomial Regression:
● If we apply a linear model on a linear dataset, then it provides us
a good result as we have seen in Simple Linear Regression, but if
es
we apply the same model without any modification on a non-linear
dataset, then it will produce a drastic output. Due to which loss
ot
function will increase, the error rate will be high, and accuracy
will be decreased.
● So for such cases, where data points are arranged in a non-linear
N
fashion, we need the Polynomial Regression model. We can
understand it in a better way using the below comparison diagram
's
of the linear dataset and non-linear dataset.
ah
hm
Re
In the above image, we have taken a dataset which is arranged
non-linearly. So if we try to cover it with a linear model, then we can
clearly see that it hardly covers any data point. On the other hand, a
curve is suitable to cover most of the data points, which is of the
Polynomial model.
Hence, if the datasets are arranged in a non-linear fashion, then we
should use the Polynomial Regression model instead of Simple Linear
Regression.
Note: A Polynomial Regression algorithm is also called Polynomial Linear
Regression because it does not depend on the variables, instead, it
depends on the coefficients, which are arranged in a linear fashion.
es
Equation of the Polynomial Regression Model:
Simple Linear Regression equation:
ot
y = b0+b1x
Polynomial Regression equation:
2 3 𝑛
N
𝑦 = 𝑏𝑜 + 𝑏
1𝑥 + 𝑏1𝑥 + 𝑏2𝑥
+ ... + 𝑏𝑛𝑥
Q13) Identify the domain where you can apply linear regression and
's
polynomial regression
ah
Domain Linear Regression Polynomial Regression
Economics - Stock price - Modeling non-linear stock
hm
and Finance prediction price trends
- Sales forecasting - Demand forecasting with
- House price non-linear growth
Re
prediction
Healthcare - Medical cost - Disease progression
prediction modeling (e.g., cancer growth)
- Predicting BMI - Drug efficacy over time
based on
height/weight
Engineering - Energy consumption - Trajectory modeling in
prediction based on physics
usage patterns - Heat transfer in complex
systems
Social - Income vs. - Modeling complex social
Sciences education level behaviors (e.g., crime rate
es
- Predicting fluctuations)
population trends
ot
Marketing - Customer spending - Customer lifetime value with
and prediction non-linear trends
N
Advertising - Pricing models - Advanced pricing models
with curve fitting
Environmental - Temperature vs. - Climate change modeling
's
Science energy consumption - Pollution level prediction
(complex interactions)
ah
Agriculture - Predicting crop - Modeling crop yield
yield based on linear considering non-linear factors
hm
factors like rainfall like soil fertility changes over
time
Re
Physics and - Simple force or - Complex motion trajectories
Mechanics speed predictions - Fatigue testing of materials
Education - Predicting student - Modeling non-linear trends
performance based in student learning behavior
on study hours
Real Estate - House price - Real estate value prediction
prediction based on based on complex, non-linear
linear factors (e.g., factors (e.g., proximity to
size, location) future development)
Q14) What is reinforcement learning? How it is different than
supervised and unsupervised learning?
es
Reinforcement Learning (RL)is a branch of machine learning focused on
ot
making decisions to maximize cumulative rewards in a given situation.
Unlike supervised learning, which relies on a training dataset with
N
predefined answers, RL involves learning through experience. In RL, an
agent learns to achieve a goal in an uncertain, potentially complex
environment by performing actions and receiving feedback through
's
rewards or penalties.
ah
Key Concepts of Reinforcement Learning
Agent: The learner or decision-maker.
hm
Environment: Everything the agent interacts with.
State: A specific situation in which the agent finds itself.
Action: All possible moves the agent can make.
Re
Reward: Feedback from the environment based on the action taken.
How Reinforcement Learning Works
RL operates on the principle of learning optimal behavior through trial
and error. The agent takes actions within the environment, receives
rewards or penalties, and adjusts its behavior to maximize the
cumulative reward. This learning process is characterized by the
following elements:
Policy: A strategy used by the agent to determine the next action
based on the current state.
Reward Function:A function that provides a scalar feedback signal
based on the state and action.
es
Value Function:A function that estimates the expected cumulative
reward from a given state.
ot
Model of the Environment: A representation of the environment that
helps in planning by predicting future states and rewards.
Example: Navigating a Maze
N
The problem is as follows: We have an agent and a reward, with many
's
hurdles in between. The agent is supposed to find the best possible
path to reach the reward. The following problem explains the problem
ah
more easily.
hm
Re
The above image shows the robot, diamond, and fire. The goal of the
robot is to get the reward that is the diamond and avoid the hurdles
that are fired. The robot learns by trying all the possible paths and
then choosing the path which gives him the reward with the least
hurdles. Each right step will give the robot a reward and each wrong
step will subtract the reward of the robot. The total reward will be
calculated when it reaches the final reward that is the diamond.
es
Main points in Reinforcement learning –
ot
● Input: The input should be an initial state from which the model
will start
● Output: There are many possible outputs as there are a variety
N
of solutions to a particular problem
● Training: The training is based upon the input, The model will
's
return a state and the user will decide to reward or punish the
model based on its output.
ah
● The model keeps continues to learn.
● The best solution is decided based on the maximum reward.
hm
Feature Supervised Unsupervised Reinforcement
Learning Learning Learning
Re
Definition Learning from Learning from Learning through
labeled data with unlabeled data, interaction with an
the correct output finding hidden environment, using
provided for each patterns or rewards and penalties.
input. structures.
Input Data Labeled data Unlabeled data (no The environment
(input-output pairs explicit output provides states and
are provided). labels). the agent chooses
actions to receive
feedback
(rewards/penalties).
Objective Predict the correct Discover hidden Maximize the
es
label for new, patterns or cumulative reward by
unseen data. groupings in the learning the best
data. sequence of actions.
ot
Learning The model is The model The agent learns
Process trained by organizes data through trial and
minimizing the
N
based on
predictions and true specific guidance.
error, receiving
difference between similarity, without feedback for its
actions and adjusting
's
labels (error). its strategy.
Data Requires large Does not require Data comes from
ah
Dependency amounts of labeled labeled data; continuous interaction
data for training. focuses on with an environment
exploring data (dynamic and
hm
structure. sequential).
Common - Linear Regression - K-Means - Q-Learning
Algorithms - Decision Trees Clustering - Deep Q-Network
Re
- Random Forests - Hierarchical (DQN)
- Support Vector Clustering - SARSA
Machines (SVMs) - Principal - Proximal Policy
Component Optimization (PPO)
Analysis (PCA)
Example - Spam detection - Customer - Robotics
Applications - Image segmentation - Game AI (e.g.,
classification - Anomaly AlphaGo)
- Medical diagnosis detection - Autonomous driving
- Market basket - Personalized
analysis recommendations
Type of Explicit feedback in No feedback; the Reward signal or
es
Feedback the form of labeled model penalty after each
data self-discovers action (delayed
(correct/incorrect patterns in the feedback).
ot
labels). data.
Task Type Classification, Clustering, Sequential
Advantages
Regression
- Provides accurate
predictions with
N
Association
- Can work with
unlabeled data,
decision-making tasks
- Learns complex
strategies and adapts
's
well-labeled data. which is more to dynamic
- Clear evaluation readily available. environments.
ah
metrics (accuracy, - Can reveal - Maximizes long-term
precision, recall, unknown patterns rewards.
etc.). in the data.
hm
Disadvantage - Requires labeled - Harder to - Requires a large
s data, which can be evaluate the amount of trial and
costly and performance error.
Re
time-consuming to without clear - May struggle with
obtain. labels. long-term planning due
- Can be less to delayed rewards.
interpretable.
Real-World - Predicting house - Grouping - Training a robot to
Example prices customers into navigate a room
- Classifying emails segments for - Self-driving cars
as spam/not spam targeted learning to avoid
marketing obstacles
- Product
recommendations
es
Q15) Compare polynomial and linear regression
ot
Feature Linear Regression Polynomial Regression
N
Definition Models the relationship Models the relationship
between the dependent between the dependent and
and independent independent variables as a
's
variables as a straight polynomial curve.
line.
ah
Equation y=b0+b1xy y=b0+b1x+b2x^2+⋯+bnx^n
hm
Type of Assumes alinear Assumes anon-linear
Relationship relationship between relationship that can be
variables. represented as a polynomial.
Re
Complexity Simple, requires fewer More complex, as
computational higher-degree polynomials
resources. increase the model’s
complexity.
Fitting Fits straight lines to Fits curves to data; better
Ability data; useful for linearly for capturing more complex,
separable data. non-linear trends.
Overfitting Lower risk of Higher risk of overfitting
Risk overfitting, especially with high-degree
for small datasets. polynomials.
es
Applications - Stock price prediction - Trajectory modeling
- Sales forecasting - Disease progression
ot
- House price prediction - Climate change modeling
- Medical cost - Crop yield prediction
N
prediction
Interpretabi Easier to interpret and Can be harder to interpret
lity understand, as the as the complexity of the
's
relationship is simple. curve increases.
ah
Handling of Best forlineardata Suitable fornon-lineardata
Data patterns where a patterns where curves
Patterns straight-line better fit the data.
hm
approximation is
sufficient.
Re
Example Predicting house prices Modeling population growth
based on features like trends or predicting complex
area, number of rooms, physics-based trajectories.
etc.
Computation More computationally Requires more computational
al Efficiency efficient and faster. power, especially for
high-degree polynomials.
Overfitting Less prone to Needs regularization
Prevention overfitting, works well techniques (e.g., Lasso or
with small data sizes. Ridge) to avoid overfitting.
es
Unit 2
ot
Q1) Define the term classification in machine learning by providing
three real life examples
N
Classification:A classification problem is when theoutput variable is a
's
category, such as “Red” or “blue” , “disease” or “no disease”.
Classification is a type of supervised learning that is used to predict
ah
categorical values, such as whether a customer will churn or not,
whether an email is spam or not, or whether a medical image shows a
tumor or not. Classification algorithms learn a function that maps from
hm
the input features to a probability distribution over the output classes.
Some common classification algorithms include:
Re
Logistic Regression, Support Vector Machines, Decision Trees, Random
Forests, Naive Baye
Evaluation Metrics of Classification:
Accuracy, Precision, Recall, F1 Score, Confusion Matrix
Advantages of Supervised Learning:
1. Since Supervised Learning works with a data set, so we can have an
exact idea about the classes of objects.
2. These algorithms are useful or helpful in predicting the output
based on the prior experience.
es
Disadvantages of Supervised Learning:
ot
1. These algorithms are not able to solve complex problems.
2. It may predict the wrong output if the test data is different from
the training data.
N
3. It requires lot of computational time to train the algorithm.
's
Applications of Supervised Learning:
1. Email Spam Detection
ah
● Application: Email services like Gmail and Outlook use
classification algorithms to automatically filter out spam emails
hm
from the inbox.
● How it Works: A classification model is trained on a labeled
dataset of emails, where each email is marked as either "spam" or
Re
"not spam." The model learns patterns from the email content,
sender information, and other metadata to classify incoming
emails.
● Classification Type: Binary classification (Spam/NotSpam).
2. Medical Diagnosis
● Application: In healthcare, classification algorithms are used to
diagnose diseases based on patient data such as symptoms, test
results, and medical history.
● How it Works: For instance, a model trained on labeledmedical
datasets can classify whether a patient has a specific disease
(e.g., diabetes, cancer) or not, based on input features like blood
sugar levels, age, weight, and more.
es
● Classification Type: Multi-class classification (e.g.,Disease A,
Disease B, or No Disease).
ot
3. Credit Card Fraud Detection
● Application: Financial institutions use classificationto detect
N
fraudulent credit card transactions.
● How it Works: A classification model is trained on past
's
transaction data, where transactions are labeled as either
"fraudulent" or "legitimate." The model learns patterns and can
ah
flag suspicious transactions for further investigation.
● Classification Type: Binary classification (Fraud/Legitimate).
hm
Q2) How one can differentiate classification and clustering.
Feature Classification Clustering
Re
Definition Assigns predefined Groups data points into
labels to data points clusters based on similarity,
based on training data. without predefined labels.
Type of Supervised Learning Unsupervised Learning
Learning (requires labeled data). (works with unlabeled data).
Objective Predict the Discover hidden patterns or
category/class for new structures in data by
data points. grouping similar items.
Input Data Labeled data (with Unlabeled data (no prior
known class labels). knowledge of classes).
Output Discrete class labels Groupings or clusters of
es
(e.g., "spam" or "not data points (e.g., cluster 1,
spam"). cluster 2).
ot
Common - Logistic Regression - K-Means Clustering
Algorithms - Decision Trees - Hierarchical Clustering
- Support Vector
Machines (SVMs) N - DBSCAN
's
Evaluation - Accuracy - Silhouette Score
Metrics - Precision - Davies-Bouldin Index
ah
- Recall - Calinski-Harabasz Score
- F1 Score
hm
Real-World - Email spam detection - Customer segmentation
Example (spam/not spam) - Image segmentation
- Disease diagnosis
Re
(disease/no disease)
Data Requires labeled data No need for labeled data; it
Dependency for training and groups data based on
classification. similarities.
Output Predicts a class or Assigns data points to
Interpretati category based on clusters based on their
on learned patterns. relative distance or
similarity.
Number of Known in advance (e.g., Number of clusters may or
Categories two classes for binary may not be known and can
es
classification). vary.
Examples of - Fraud detection - Market segmentation
ot
Use - Medical diagnosis - Social network analysis
- Sentiment analysis - Anomaly detection
Advantages
results
- Can handle complex
N
- Clear, interpretable - Does not require labeled
data
- Can reveal hidden patterns
's
labeled data in the data
ah
Disadvantage - Requires labeled data, - Harder to interpret
s which can be costly to results
obtain - Sensitive to noise and
hm
- Not effective for data distribution
discovering unknown
patterns
Re
Q3) Explain the working of random forest by giving example
Random Forest is a popular machine learning algorithm that belongs to
the supervised learning technique. It can be used for both
Classification and Regression problems in ML. It is based on the
concept of ensemble learning, which is a process of combining multiple
classifiers to solve a complex problem and to improve the performance
of the model.
As the name suggests, "Random Forest is a classifier that contains a
number of decision trees on various subsets of the given dataset and
takes the average to improve the predictive accuracy of that dataset."
es
Instead of relying on one decision tree, the random forest takes the
prediction from each tree and based on the majority votes of
ot
predictions, and it predicts the final output.
The greater number of trees in the forest leads to higher accuracy
N
and prevents the problem of overfitting.
's
The below diagram explains the working of the Random Forest
algorithm:
ah
hm
Re
There are two assumptions for a better Random forest classifier:
1. There should be some actual values in the feature variable of the
dataset so that the classifier can predict accurate results rather
than a guessed result.
2. The predictions from each tree must have very low correlations.
Why use Random Forest?
es
● It takes less training time as compared to other algorithms.
● It predicts output with high accuracy, even for the large dataset
ot
it runs efficiently.
● It can also maintain accuracy when a large proportion of data is
missing.
N
How does Random Forest algorithm work?
's
Random Forest works in two-phase first is to create the random forest
by combining N decision tree, and second is to make predictions for
ah
each tree created in the first phase.
Step-1: Select random K data points from the training set.
hm
Step-2: Build the decision trees associated with the selected data
points (Subsets).
Re
Step-3: Choose the number N for decision trees that you want to build.
Step-4: Repeat Step 1 & 2.
Step-5: For new data points, find the predictions of each decision tree,
and assign the new data points to the category that wins the majority
votes.
Example: Suppose there is a dataset that contains multiple fruit
images. So, this dataset is given to the Random forest classifier. The
dataset is divided into subsets and given to each decision tree. During
es
the training phase, each decision tree produces a prediction result, and
when a new data point occurs, then based on the majority of results,
ot
the Random Forest classifier predicts the final decision. Consider the
below image:
N
's
ah
hm
Re
Applications of Random Forest
● Banking: Banking sector mostly uses this algorithm for the
identification of loan risk.
● Medicine: With the help of this algorithm, disease trends and
risks of the disease can be identified.
● Land Use: We can identify the areas of similar land use by this
es
algorithm.
● Marketing: Marketing trends can be identified using this
ot
algorithm.
Advantages of Random Forest
N
● Random Forest is capable of performing both Classification and
Regression tasks.
's
● It is capable of handling large datasets with high dimensionality.
● It enhances the accuracy of the model and prevents the
ah
overfitting issue.
Disadvantages of Random Forest
hm
● Although random forest can be used for both classification and
regression tasks, it is not more suitable for Regression tasks.
Re
Q4) List and elaborate applications of random forest
1. Healthcare: Disease Diagnosis and Risk Prediction
● Application: Random Forest is extensively used inthe medical
field to predict diseases, identify high-risk patients, and assist in
diagnostic procedures.
● How it Works: The algorithm processes medical data(such as
patient history, symptoms, test results) to classify whether a
patient is at risk of a disease (e.g., diabetes, heart disease). It
also helps predict outcomes like the chances of recovery based on
multiple medical variables.
● Example: Predicting whether a patient has cancer basedon biopsy
features, identifying high-risk cardiovascular patients, or
es
predicting whether a person is prone to certain genetic diseases.
2. Finance: Fraud Detection
ot
● Application: Random Forest is widely used to detect fraudulent
transactions in real-time within financial institutions, such as
banks or credit card companies.
N
● How it Works: The algorithm analyzes past transactiondata
's
labeled as "fraudulent" or "legitimate" and learns patterns that
indicate fraud. It can then flag suspicious transactions for
ah
further investigation.
● Example: Identifying credit card fraud by analyzingtransaction
behaviors (e.g., location, transaction time, amount), or predicting
hm
fraudulent insurance claims.
3. Marketing: Customer Segmentation and Recommendation Systems
Re
● Application: In marketing, Random Forest is used forcustomer
segmentation, personalized recommendations, and identifying
customer churn.
● How it Works: By analyzing customer behaviors, purchasing
history, and demographic information, the algorithm can classify
customers into segments and help marketers create targeted
campaigns.
● Example: Grouping customers with similar buying patterns,
predicting which customers are likely to churn, or recommending
products based on past purchases.
4. E-commerce: Product Recommendations
es
● Application: Random Forest can be used in e-commerceplatforms
to provide personalized product recommendations.
ot
● How it Works: The algorithm analyzes past purchase behavior,
browsing history, and customer preferences to suggest relevant
products.
N
● Example: Amazon’s recommendation engine uses RandomForest
models to suggest products that customers might want to buy
's
based on past behavior and similar user profiles.
5. Banking: Credit Risk Analysis
ah
● Application: Banks use Random Forest for assessing
creditworthiness and determining whether to approve loan
hm
applications or credit card limits.
● How it Works: The model evaluates financial data,credit history,
income, and other features to classify whether a borrower is
Re
likely to default on a loan or manage credit well.
● Example: Predicting the risk level of a loan applicantbased on
their credit score, employment history, and debt-to-income ratio.
6. Agriculture: Crop Disease Detection and Yield Prediction
● Application: In agriculture, Random Forest is used for crop
disease identification and yield prediction.
● How it Works: The model can analyze features likesoil
conditions, temperature, rainfall, and satellite images to classify
whether crops are healthy or diseased and to predict crop yields
for the season.
● Example: Identifying diseased crops from image data,predicting
es
wheat yield based on climate and soil data.
7. Natural Language Processing (NLP): Sentiment Analysis
ot
● Application: Random Forest is employed in sentiment analysis to
classify text (such as reviews, social media posts) into categories
N
like positive, negative, or neutral sentiment.
● How it Works: The algorithm analyzes word frequencies,
's
sentence structures, and other textual features to classify text
into various sentiment categories.
ah
● Example: Classifying movie reviews, product feedback,or social
media posts as positive, negative, or neutral for brand reputation
analysis.
hm
8. Cybersecurity: Intrusion Detection
● Application: Random Forest is used to detect networkintrusions
Re
and cybersecurity threats by identifying abnormal patterns in
network traffic.
● How it Works: By learning from past intrusion data,the model
can classify incoming traffic as normal or malicious, helping
network administrators flag suspicious activities.
● Example: Detecting unauthorized access attempts, malware
attacks, or unusual login patterns in a network.
9. Environmental Science: Climate Change Prediction
● Application: Random Forest is employed to predictclimate
patterns and understand environmental changes based on vast
amounts of climate data.
es
● How it Works: The algorithm processes historical weatherdata,
temperature records, CO2 levels, and other environmental
ot
factors to predict future climate scenarios.
● Example: Predicting temperature rise, rainfall patterns, or CO2
levels in the atmosphere for the next decade based on historical
data.
N
10. Manufacturing: Quality Control and Fault Detection
's
● Application: Random Forest is used to improve productquality and
ah
detect defects in manufacturing processes.
● How it Works: The algorithm analyzes data from manufacturing
equipment, such as sensor readings, production rates, and
hm
material properties, to classify whether a product meets quality
standards or if there is a defect.
● Example: Detecting faulty parts in a car manufacturingprocess
Re
by analyzing machine sensor data, or predicting machinery
breakdowns based on operational data.
11. Bioinformatics: Gene Classification
● Application: In bioinformatics, Random Forest is applied to
classify genes based on their expression profiles or identify gene
mutations linked to specific diseases.
● How it Works: The algorithm analyzes genetic data,including
gene expression levels or DNA sequences, to classify genes into
categories (e.g., normal vs. mutated) or predict the functions of
unknown genes.
es
● Example: Classifying tumor vs. non-tumor genes basedon
expression data, identifying genetic markers associated with
ot
hereditary diseases.
12. Image Recognition: Object Detection and Classification
N
● Application: Random Forest is used in image recognition to
classify and detect objects in images.
's
● How it Works: The algorithm processes pixel values,colors,
shapes, and textures from images to classify objects or detect
ah
specific patterns.
● Example: Recognizing objects like cars, animals, orfaces in
images, or classifying handwritten digits for automated data
hm
entry.
Q5) What is confusion matrix? Provide examples
Re
What is a Confusion Matrix
The confusion matrix shows the ways in which your classification model
is confused when it makes predictions.
A Confusion matrix is an N x N matrix used for evaluating the
performance of a classification model, where N is the number of target
classes.
The matrix compares the actual target values with those predicted by
the machine learning model.
This gives us a holistic view of how well our classification model is
performing and what kinds of errors it is making.
How to Calculate a Confusion Matrix
1. You need a test dataset or a validation dataset with expected
es
outcome values.
2. Make a prediction for each row in your test dataset.
ot
3. From the expected outcomes and predictions count: The number of
correct predictions for each class.
The number of incorrect predictions for each class, organized by the
class that was predicted.
N
4. These numbers are then organized into a table, or a matrix as
's
follows:
ah
hm
Expected down the side: Each row of the matrix corresponds to a
Re
predicted class.
Predicted across the top: Each column of the matrix corresponds to
an actual class.
Confusion Matrix
True Positive (TP)
The predicted value matches the actual value
es
The actual value was positive and the model predicted a positive value
True Negative (TN)
ot
The predicted value matches the actual value
The actual value was negative and the model predicted a negative value
False Positive (FP)
N
The predicted value was falsely predicted
The actual value was negative but the model predicted a positive value
's
False Negative (FN)
The predicted value was falsely predicted
ah
The actual value was positive but the model predicted a negative value
Need for Confusion Matrix in Machine learning
hm
It evaluates the performance of the classification models, when they
make predictions on test data, and tells how good our classification
model is.
Re
• It not only tells the error made by the classifiers but also the type
of errors such as it is either type-l or type-ll error.
• With the help of the confusion matrix, we can calculate the different
parameters for the model, such as accuracy, precision, etc.
Example:
Expected Predicted
man woman
man man
woman woman
man man
es
women man
women women
ot
women women
man man
man
N
women
's
women women
ah
men classified as men: 3 women classified as women: 4
men classified as women: 2 woman classified as men: 1
hm
man woman
man 3 2
woman 1 4
Re
The total actual men in the dataset is the sum of the values on the men
column (3+2)
The total actual women in the dataset is the sum of values in the
women column (1 +4).
The correct values are organized in a diagonal line from top left to
bottom-right of the matrix (3+4).
More errors were made by predicting men as women than predicting
women as men
es
ot
True Positive:
Interpretation: You predicted positive and it's true. You predicted
N
that a woman is pregnant and she actually is.
True Negative:
's
Interpretation: You predicted negative and it's true. You predicted
that a man is not pregnant and he actually is not
ah
False Positive:
Interpretation: You predicted positive and it's false. You predicted
that a man is pregnant but he actually is not.
hm
False Negative:
Interpretation: You predicted negative and it's false. You predicted
that a woman is not pregnant but she actually is.
Re
Q6) Explain the concept of type 1 and type 2 errors by giving suitable
examples.
Confusion Matrix
True Positive (TP)
The predicted value matches the actual value
es
The actual value was positive and the model predicted a positive value
True Negative (TN)
ot
The predicted value matches the actual value
The actual value was negative and the model predicted a negative value
False Positive (FP)
N
The predicted value was falsely predicted
The actual value was negative but the model predicted a positive value
's
False Negative (FN)
The predicted value was falsely predicted
ah
The actual value was positive but the model predicted a negative value
Type 1 and Type 2 Error
hm
Scenario 1:We don't have a kitten among the group. Yet, ML algo
predicts it is there. If we accept the ML algo prediction then it is Type
1 error also known as 'False Positive'
Re
Scenario 2:We have a kitten among the group. Yet, ML algo predicts it
is not there. If we accept the ML algo prediction then it is Type 2
error also known as 'False Negative'.
Use cases of Type 1 and Type 2
Scenario/Problem Statement 1:Providing access to an asset post a
biometric scan.
Type I error: Possibility of rejection even with an authorized match.
Type II error: Possibility of acceptance even with a unauthorized
match.
Scenario/Problem Statement 2:Construction Model of a bridge is
correct
Type I error: Predicting that the model is correct when it is not.
Type II error: Predicting that a model is not correct when it is
es
correct.
Scenario/Problem Statement 3:Medical trials for a drug which is a
ot
cure for Cancer
Type I error: Predicting that a cure is found when it is not the case.
Type II error: Predicting that a cure is not found when in fact it is the
case.
N
's
Q7) Discuss following by giving suitable examples.
● Overfitting ● Underfitting
ah
Underfitting in Machine Learning
A statistical model or a machine learning algorithm is said to have
hm
underfitting when a model is too simple to capture data complexities.
It represents the inability of the model to learn the training data
effectively result in poor performance both on the training and testing
Re
data. It mainly happens when we uses very simple model with overly
simplified assumptions. To address underfitting problem of the model,
we need to use more complex models, with enhanced feature
representation, and less regularization.
Note: The underfitting model has High bias and low variance.
Reasons for Underfitting
● The model is too simple, So it may be not capable to represent
the complexities in the data.
● The input features which is used to train the model is not the
adequate representations of underlying factors influencing the
target variable.
● The size of the training dataset used is not enough.
es
● Excessive regularization are used to prevent the overfitting,
which constraint the model to capture the data well.
ot
● Features are not scaled.
Techniques to Reduce Underfitting
● Increase model complexity.
N
● Increase the number of features, performing feature
's
engineering.
● Remove noise from the data.
ah
● Increase the number of epochs or increase the duration of
training to get better results.
hm
Overfitting in Machine Learning
A statistical model is said to be overfitted when the model does not
make accurate predictions on testing data. When a model gets trained
Re
with so much data, it starts learning from the noise and inaccurate
data entries in our data set. And when testing with test data results in
High variance. Then the model does not categorize the data correctly,
because of too many details and noise. The causes of overfitting are
the non-parametric and non-linear methods because these types of
machine learning algorithms have more freedom in building the model
based on the dataset and therefore they can really build unrealistic
models. A solution to avoid overfitting is using a linear algorithm if we
have linear data or using the parameters like the maximal depth if we
are using decision trees.
In a nutshell, Overfitting is a problem where the evaluation of machine
learning algorithms on training data is different from unseen data.
es
Reasons for Overfitting:
● High variance and low bias.
ot
● The model is too complex.
● The size of the training data.
Techniques to Reduce Overfitting
N
● Improving the quality of training data reduces overfitting by
's
focusing on meaningful patterns, mitigate the risk of fitting the
noise or irrelevant features.
ah
● Increase the training data can improve the model’s ability to
generalize to unseen data and reduce the likelihood of
overfitting.
hm
● Reduce model complexity.
● Early stopping during the training phase (have an eye over the
loss over the training period as soon as loss begins to increase
Re
stop training).
● Ridge Regularization and Lasso Regularization.
● Use dropout for neural networks to tackle overfitting.
es
ot
N
's
ah
Q8) Define or explain following terms
hm
● Entropy ● Information gain
What is Entropy in Machine Learning
Re
Entropy is the measurement of disorder or impurities in the
information processed in machine learning. It determines how a
decision tree chooses to split data.
We can understand the term entropy with any simple example: flipping
a coin. When we flip a coin, then there can be two outcomes. However,
it is difficult to conclude what would be the exact outcome while
flipping a coin because there is no direct relation between flipping a
coin and its outcomes. There is a 50% probability of both outcomes;
then, in such scenarios, entropy would be high. This is the essence of
entropy in machine learning.
Mathematical Formula for Entropy
Consider a data set having a total number of N classes, then the
es
entropy (E) can be determined with the formula below:
ot
Where;
Pi = Probability of randomly selecting an example in class I;
N
Entropy always lies between 0 and 1, however depending on the number
of classes in the dataset, it can be greater than 1. But the high value
's
of
ah
Let's understand it with an example where we have a dataset having
three colors of fruits as red, green, and yellow. Suppose we have 2 red,
hm
2 green, and 4 yellow observations throughout the dataset. Then as per
the above equation:
Re
Where;
Pr = Probability of choosing red fruits;
Pg = Probability of choosing green fruits and;
Py = Probability of choosing yellow fruits.
Pr = 2/8 =1/4 [As only 2 out of 8 datasets represents red fruits]
Pg = 2/8 =1/4 [As only 2 out of 8 datasets represents green fruits]
Py = 4/8 = 1/2 [As only 4 out of 8 datasets represents yellow fruits]
Now our final equation will be such as;
es
ot
So, entropy will be 1.5.
N
's
Let's consider a case when all observations belong to the same class;
ah
then entropy will always be 0.
hm
E=−(1log21)
= 0
Re
When entropy becomes 0, then the dataset has no impurity. Datasets
with 0 impurities are not useful for learning. Further, if the entropy is
1, then this kind of dataset is good for learning.
es
What is the information gain in Entropy?
Information gain is defined as the pattern observed in the dataset and
ot
reduction in the entropy.
formula:
N
Mathematically, information gain can be expressed with the below
Information Gain = (Entropy of parent node)-(Entropy of child node)
's
Note: Information gain is calculated as 1-Entropy.
ah
Let's understand it with an example having three scenarios as follows:
hm
Re
Let's say we have a tree with a total of four values at the root node
that is split into the first level having one value in one branch (say,
Branch 1) and three values in the other branch (Branch 2). The entropy
at the root node is 1.
Now, to compute the entropy at the child node 1, the weights are taken
as ? for Branch 1 and ? for Branch 2 and are calculated using Shannon's
entropy formula. As we had seen above, the entropy for child node 2 is
zero because there is only one value in that child node, meaning there
es
is no uncertainty, and hence, the heterogeneity is not present.
ot
H(X) = - [(1/3 * log2 (1/3)) + (2/3 * log2 (2/3))] = 0.9184
The information gain for the above case is the reduction in the
weighted average of the entropy.
N
's
Information Gain = 1 - ( ¾ * 0.9184) - (¼ *0) = 0.3112
ah
The more the entropy is removed, the greater the information gain.
The higher the information gain, the better the split.
hm
Q9) Explain naïve bayes classification
● Naïve Bayes algorithm is a supervised learning algorithm, which is
Re
based on Bayes theorem and used for solving classification
problems.
● It is mainly used in text classification that includes a
high-dimensional training dataset.
● Naïve Bayes Classifier is one of the simple and most effective
Classification algorithms which helps in building the fast machine
learning models that can make quick predictions.
● It is a probabilistic classifier, which means it predicts on the
basis of the probability of an object.
● Some popular examples of Naïve Bayes Algorithm are spam
filtration, Sentimental analysis, and classifying articles.
● Naïve: It is called Naïve because it assumes that the occurrence
of a certain feature is independent of the occurrence of other
features.
es
● Such as if the fruit is identified on the bases of color, shape, and
taste, then red, spherical, and sweet fruit is recognized as an
ot
apple. Hence each feature individually contributes to identify
that it is an apple without depending on each other.
● Bayes: It is called Bayes because it depends on the principle of
Bayes' Theorem.
N
● Bayes' Theorem: Bayes' theorem is also known as Bayes' Rule or
's
Bayes' law, which is used to determine the probability of a
hypothesis with prior knowledge. It depends on the conditional
ah
probability.
● The formula for Bayes' theorem is given as:
hm
Where,
P(A|B) is Posterior probability: Probability of hypothesis A on the
Re
observed event B.
P(B|A) is Likelihood probability: Probability of the evidence given that
the probability of a hypothesis is true.
P(A) is Prior Probability: Probability of hypothesis before observing the
evidence.
P(B) is Marginal Probability: Probability of Evidence.
Advantages of Naïve Bayes Classifier:
● Naïve Bayes is one of the fast and easy ML algorithms to predict
a class of datasets.
● It can be used for Binary as well as Multi-class Classifications.
● It performs well in Multi-class predictions as compared to the
other Algorithms.
● It is the most popular choice for text classification problems.
es
Disadvantages of Naïve Bayes Classifier:
ot
● Naive Bayes assumes that all features are independent or
unrelated, so it cannot learn the relationship between features.
N
Applications of Naïve Bayes Classifier:
● It is used for Credit Scoring.
's
● It is used in medical data classification.
● It can be used in real-time predictions because Naïve Bayes
ah
Classifier is an eager learner.
● It is used in Text classification such as Spam filtering and
Sentiment analysis.
hm
Types of Naïve Bayes Model:
● Gaussian: The Gaussian model assumes that features follow a
Re
normal distribution. This means if predictors take continuous
values instead of discrete, then the model assumes that these
values are sampled from the Gaussian distribution.
● Multinomial: The Multinomial Naïve Bayes classifier is used when
the data is multinomial distributed. It is primarily used for
document classification problems, it means a particular document
belongs to which category such as Sports, Politics, education, etc.
The classifier uses the frequency of words for the predictors.
● Bernoulli: The Bernoulli classifier works similar to the Multinomial
classifier, but the predictor variables are the independent
Booleans variables. Such as if a particular word is present or not
in a document. This model is also famous for document
classification tasks.
es
Q10) Describe advantages and applications of naïve bayes
ot
classification
Advantages of Naïve Bayes Classification
N
1. Simplicity and Ease of Implementation:
○ Naïve Bayes is simple and easy to implement. It assumes
's
independence between the features, which reduces
complexity, making it suitable for quick applications with
ah
limited computational resources.
2. Fast and Efficient:
○ Naïve Bayes is computationally efficient and works well with
hm
large datasets. Its training and prediction times are fast
because it simplifies probability calculations using
conditional independence.
Re
3. Works Well with Small Datasets:
○ Despite being a simple algorithm, Naïve Bayes performs
surprisingly well even when the dataset is small. This makes
it ideal in situations where gathering large amounts of data
is difficult.
4. Performs Well with Categorical Data:
○ Naïve Bayes works particularly well when the input features
are categorical (e.g., for text classification problems). It
can handle both binary and multi-class classification tasks
effectively.
5. Performs Well for Multiclass Classification:
○ Unlike some algorithms that struggle with multiclass
classification, Naïve Bayes handles multiple classes very
es
well. This makes it ideal for tasks with more than two
outcomes.
ot
6. Robust to Irrelevant Features:
○ Naïve Bayes is relatively immune to irrelevant features in
the data. Even if the assumption of independence between
N
features is violated, it can still perform well in many
practical applications.
's
7. Performs Well with Text Data and Natural Language
Processing (NLP):
ah
○ Naïve Bayes is popular in text-related tasks (e.g., spam
filtering, sentiment analysis) because of its ability to handle
high-dimensional data and its efficiency with text
hm
classification tasks.
8. Handles Missing Data:
○ Naïve Bayes can handle missing data relatively well. While
Re
some machine learning algorithms may need data imputation
methods, Naïve Bayes can make predictions even with
missing attributes by ignoring them during probability
calculations.
Applications of Naïve Bayes Classification
1. Spam Filtering:
○ Application: Email service providers like Gmail andYahoo use
Naïve Bayes for spam detection. The classifier labels emails
as either "spam" or "not spam" based on their content,
sender information, and other features.
○ How it Works: The algorithm is trained on a datasetof
labeled emails (spam and not spam). It calculates the
es
likelihood of an email being spam based on the presence or
absence of certain keywords and features.
ot
2. Sentiment Analysis:
○ Application: Naïve Bayes is used to classify customer
reviews, social media posts, or feedback into categories like
N
positive, negative, or neutral sentiment.
○ How it Works: By analyzing the frequency of positiveor
's
negative words in a dataset of labeled text (reviews or
tweets), Naïve Bayes can predict the sentiment of new text
ah
data.
○ Example: It’s widely used in e-commerce platforms to
analyze customer reviews and gauge the overall sentiment
hm
towards products.
3. Document Classification:
○ Application: Naïve Bayes is widely used in text classification
Re
tasks such as news categorization, topic labeling, and
document classification.
○ How it Works: The classifier analyzes words or phrasesin
documents and classifies them into predefined categories,
such as politics, sports, entertainment, or technology.
○ Example: News websites use Naïve Bayes to automatically
categorize articles based on their content.
4. Medical Diagnosis:
○ Application: Naïve Bayes is used in healthcare topredict
diseases based on patient data such as symptoms, medical
history, and test results.
○ How it Works: The algorithm is trained on a datasetof
patient data with known diagnoses. It then uses this
information to predict whether new patients might have a
es
particular disease based on the likelihood of specific
symptoms.
ot
○ Example: Predicting the likelihood of a patient having a
disease like diabetes or heart disease based on input
features like age, weight, blood sugar levels, etc.
5. Recommendation Systems:
N
○ Application: Naïve Bayes is applied in recommendation
's
engines to suggest items such as movies, books, or products
to users based on their preferences.
ah
○ How it Works: By analyzing user behavior and preferences
(such as past purchases or movie ratings), the algorithm
classifies items into categories (e.g., “highly recommended”
hm
or “not recommended”) and makes personalized
recommendations.
○ Example: Netflix or Amazon recommending movies or
Re
products based on user preferences.
6. Credit Scoring and Risk Prediction:
○ Application: Banks and financial institutions useNaïve Bayes
to assess the creditworthiness of loan applicants and
predict the risk of default.
○ How it Works: The algorithm analyzes features such as
credit history, income, and employment to classify
customers into low-risk or high-risk categories.
○ Example: Predicting whether a customer is likely todefault
on a loan or not, based on their financial behavior.
7. Face Recognition:
○ Application: Naïve Bayes can be used in facial recognition
es
systems to classify faces in images or videos.
○ How it Works: The algorithm analyzes facial features,such
ot
as distance between eyes, shape of the nose, etc., and
matches them to pre-classified images in the database.
○ Example: Used in security systems to recognize andverify
individuals' identities.
8. Anomaly Detection: N
's
○ Application: Naïve Bayes is applied in cybersecurityto
detect unusual patterns or anomalies, such as fraud or
ah
network intrusions.
○ How it Works: It learns the normal behavior from historical
data and flags any outliers or anomalies as potential threats.
hm
○ Example: Detecting unusual login attempts or financial
transactions that might indicate fraud.
9. Real-Time Prediction in E-commerce:
Re
○ Application: Naïve Bayes is used for real-time predictions,
such as determining whether a customer will make a
purchase or abandon the cart.
○ How it Works: By analyzing user behavior data, the
algorithm classifies customers into groups, such as “likely to
purchase” or “unlikely to purchase,” in real time.
○ Example: E-commerce sites like Amazon may use this to
offer last-minute discounts to users who are likely to
abandon their shopping cart.
Q11) Problems based on decision tree CART/ ID3
Q12) Problems based on naïve bayes
es
Numerical PDF
ot
Q13) Explain the working of SVM
SUPPORT VECTOR MACHINE
N
SVM is a method for classification of both linear and non-linear data.
's
Linearly Separable Data:
If the given data is classified into distinct classes such that
ah
they can be separated by adecision boundary, it is called as
Linearly Separable Data
hm
If the given data is classified into distinct classes such that
they cannot be separated by a decision boundary, it is called
Non-linearly Separable Data. Since it cannot be separated by a
Re
single line, it is non-linear.
SVM uses the concept of MMH (Maximum Marginal HyperPlane)
The goal of the SVM algorithm is to create the best line or
decision boundary that can segregate ‘n’ dimensional space into
classes so that we can easily put the new data points in the
correct category in the future.
This best-decision boundary is known asHyperplane
SVM chooses the extreme points/vectors that helps in creating
the Hyperplane.
es
These extreme cases are called Support Vectors and hence the
algorithm is termed as Support Vector Machine(SVM).
ot
The line formed by joining the points closest to the hyperplane
is the Margin.
N
Margin is the distance between the support vectors and the
hyperplane.
's
TERMINOLOGIES:
1. Hyperplane:It is a decision boundary used to separate
ah
data points of different classes.
For a linear classification, it will be a linear equation:
hm
𝑊𝑥 + 𝑏
= 0
Re
where,
W = weight vector
b = bias
We can write the equation for the two classes:
𝑊𝑥 + 𝑏
≥1
𝑓
𝑜𝑟𝑦
𝑖 = 1
𝑊𝑥 + 𝑏
≥1
𝑓
𝑜𝑟𝑦
𝑖 = − 1
Considering these two equalities,
𝑦(𝑊𝑥 + 𝑏
) = 1
This is used to decide the Support Vectors.
2.These are the closest data points to the hyperplane which
es
plays a critical role in deciding the hyperplane and margin.
3.Margins are of two types : Hard margin & Soft Margin
ot
N
's
ah
hm
Hard Margin : The maximum margin hyperplane or the hard
margin is a hyperplane that properly separates the data points
Re
of different categories without any miss-classifications.
Soft Margin : When the data is not perfectly separable or
contain outliers, SVM permits a soft margin technique.
4.Each datapoint has a Slack Variableintroducedbythe
soft-margin formulation, which softens thestrictmargin
requirements and permits certain miss-classifications or
violations.
5. The margin is calculated as :
es
ot
Types of SVM:
1. Linear SVM
2. Non-Linear SVM
N
's
Advantages of SVM:
1.Effective in high-dimensional cases.
ah
2.Different kernel function can be specified for the different
functions as it is possible to specify custom kernel.
hm
3.It is memory efficient.
Disadvantages:
Re
1.If the number of features is much greater than the number
of samples, avoid overfeeding in choosing the kernel
functions.
2.SVM’s do not directly provide probability estimates , these
are calculated using an expensive Five-Fold Cross-Validation.
Q14) Applications of SVM
1. Image Classification
● Application: SVM is widely used for classifying imagesin
computer vision tasks, such as facial recognition, object
detection, and handwriting recognition.
● How it Works: The algorithm can effectively classifyimages by
es
finding the optimal hyperplane that separates different classes in
the feature space derived from the image data.
ot
● Example: Recognizing handwritten digits in the MNIST dataset
or classifying images of cats and dogs.
2. Text Classification
N
● Application: SVM is employed for text categorizationtasks, such
's
as spam detection, sentiment analysis, and document
classification.
ah
● How it Works: The algorithm converts text data intonumerical
feature vectors using techniques like TF-IDF or word embeddings
and then classifies the documents based on these features.
hm
● Example: Classifying emails as spam or non-spam, ordetermining
the sentiment of product reviews as positive or negative.
Re
3. Bioinformatics
● Application: In bioinformatics, SVM is used for classifyinggenes,
proteins, and biological sequences based on their features.
● How it Works: It helps in identifying gene functionsor
predicting protein structures by analyzing complex biological
data.
● Example: Classifying genes associated with particular diseases or
predicting protein-protein interactions.
4. Finance
● Application: SVM is applied in financial markets forcredit
scoring, fraud detection, and stock price prediction.
● How it Works: It analyzes historical financial datato classify
es
transactions as fraudulent or legitimate, or to predict whether a
stock price will rise or fall.
ot
● Example: Classifying credit applicants into "approved"or "denied"
categories based on their financial history.
5. Medical Diagnosis
N
● Application: SVM is used to assist in diagnosing diseasesbased on
's
patient data, such as symptoms and medical history.
● How it Works: By analyzing various features relatedto patient
ah
health, SVM can classify individuals as healthy or as having a
particular disease.
● Example: Diagnosing diseases such as cancer by analyzingmedical
hm
imaging data or patient biomarkers.
6. Face Detection and Recognition
Re
● Application: SVM is used in computer vision for facedetection
and recognition in images and videos.
● How it Works: The algorithm classifies regions inan image as
containing a face or not, based on features extracted from the
image.
● Example: Implementing face recognition systems in security
applications or social media platforms.
7. Customer Segmentation
● Application: SVM can be used for customer segmentationin
marketing to classify customers into different groups based on
purchasing behavior and preferences.
es
● How it Works: By analyzing customer data, SVM identifies
distinct groups, allowing marketers to target specific segments
ot
with tailored campaigns.
● Example: Classifying customers as "high value," "lowvalue," or "at
risk" based on their purchasing history.
8. Anomaly Detection N
's
● Application: SVM is employed for anomaly detectiontasks in
various fields, including cybersecurity, fraud detection, and
ah
network security.
● How it Works: The algorithm can identify unusual patternsor
outliers in data, classifying them as anomalies that may require
hm
further investigation.
● Example: Detecting fraudulent transactions in creditcard
processing or identifying potential intrusions in network traffic.
Re
9. Natural Language Processing (NLP)
● Application: SVM is used in NLP tasks, such as part-of-speech
tagging, named entity recognition, and language identification.
● How it Works: It classifies words or phrases based on their
contextual features to determine their role or identity within
text.
● Example: Classifying sentences as declarative, interrogative,or
exclamatory based on their structure.
10. Time Series Forecasting
es
● Application: SVM can be utilized for forecasting timeseries data
in fields like economics, weather prediction, and stock market
ot
analysis.
● How it Works: By analyzing historical data trends,SVM can
predict future values in a time series dataset.
N
● Example: Predicting future stock prices based on historical
trends or forecasting weather conditions based on past climate
's
data.
11. Robotics
ah
● Application: SVM is applied in robotics for objectrecognition,
navigation, and human-robot interaction.
hm
● How it Works: It helps robots classify objects intheir
environment and make decisions based on the classified data.
● Example: Enabling robots to recognize and pick upspecific
Re
objects in an industrial setting.
12. Environmental Science
● Application: SVM is used in environmental monitoringand
classification of various environmental data, such as land cover
classification and species distribution modeling.
● How it Works: By analyzing satellite imagery and ecological data,
SVM can classify land types and predict species habitats.
● Example: Classifying land use types (e.g., urban,agricultural,
forest) from satellite images.
Q15) what do you mean by hypothesis? Provide examples for null
and alternate hypothesis with explanation. Provide working of null
and alternate hypothesis.
es
What is Hypothesis Testing?
ot
Hypothesis testing is a statistical method that is used in making
statistical decisions using experimental data.
population parameter.
Ex:
N
Hypothesis Testing is basically an assumption that we make about the
's
1) you say an average student in class is 40 or a boy is taller than girls.
2) Some scientists claim that ultraviolet (UV) light can damage the
ah
eyes then it may also cause blindness.
hm
Re
Terms
Hypothesis space (H):Hypothesis space is defined as a set of all
possible legal hypotheses; hence it is also known as a hypothesis set
Hypothesis (h):It is defined as the approximate function that best
describes the target in supervised machine learning algorithms. It is
primarily based on data as well as bias and restrictions applied to data.
es
ot
N
's
ah
hm
Need of Hypothesis
Hypothesis testing is an essential procedure in statistics.
A hypothesis test evaluates two mutually exclusive statements about a
Re
population to determine which statement is best supported by the
sample data. When we say that a finding is statistically significant
means a hypothesis test.
If a person gets 7 hours of sleep, then he will feel less fatigue than if
he sleeps less. Consumption of sugary drinks every day leads to obesity
Parameters of Hypothesis Testing:
Null Hypothesis
Alternate Hypothesis
Parameter Null Hypothesis Alternate Hypothesis
Definition null hypothesis is a
A n alternative
A
statement in which hypothesis is a
es
there is no relation statement in which
between the two there is some
ot
Variables. statistical
relationship between
the two variables.
What it is?
N
enerally,
G
researchers try to
esearchers try to
R
accept or
's
reject or disprove it prove it.
Testing Process Indirect and Implicit Direct and Explicit
ah
P-Value ull hypothesis is
N n alternative
A
rejected if hypothesis is
hm
the p-value is less accepted if the
than the alpha-value; p-value is less than
otherwise, it is the alpha-value
accepted. otherwise, it is
Re
rejected.
Notation H0 H1
Symbol Used quality Symbols =,
E nequality Symbols !=,
I
<=, >= !<=, !>=
Effect on Bio-fertilizer ‘x’ increases plant growth
Alternative Hypothesis H1:Application of Bio-fertilizer ‘x’ increases
plant growth
Null Hypothesis H1:Application of Bio-fertilizer ‘x’ does not increase
plant growth
Q16) Write a short note on Multivariate Regression
es
Multivariate Regression
ot
Multivariate regression is a statistical technique that uses a
mathematical model to estimate the relationship between a dependent
variable and multiple independent variables
N
It's an extension of linear regression, which only involves one response
's
variable. Multivariate regression can be used in a variety of
applications, including: Identifying risk factors for an outcome,
ah
Determining the effect of a procedure on an outcome, Comparing
different treatment strategies, Quantifying the magnitude of an
effect, and Developing risk-prediction models.
hm
Multivariate Regression is a method used to measure the degree at
which more than one independent variable (predictors) and more than
Re
one dependent variable (responses), are linearly related. The method is
broadly used to predict the behavior of the response variables
associated to changes in the predictor variables, once a desired degree
of relation has been established.
The Multivariate Regression model, relates more than one predictor
and more than one response.
Y = X*B+ ϵ
Here are some examples of how multivariate regression can be used:
es
● Pesticide concentration in surface water
A multivariate regression model can estimate the relationship between
ot
river flow and seasonal pesticide use, and how these factors affect
pesticide concentration in surface water.
● Intracranial bleeding
N
A multivariate logistic regression analysis can identify the strongest
predictors of intracranial bleeding, such as vomiting/nausea and
's
seizures.
● Multiple genetic variants and neuroimaging phenotypes
ah
A multivariate regression model can capture the complex relationships
between genes and brain measurements.
hm
Q17) State the importance of feature selection. How it is useful in
machine learning algorithms?
Re
Feature selection is a critical step in the machine learning pipeline that
involves selecting a subset of relevant features (or variables) for use in
model construction. It plays a significant role in improving the
performance of machine learning algorithms. Here’s an overview of its
importance and usefulness:
Importance of Feature Selection
1. Reduces Overfitting:
○ By eliminating irrelevant or redundant features, feature
selection helps to reduce the complexity of the model. This,
in turn, lowers the risk of overfitting, where the model
learns noise instead of the underlying patterns in the
training data.
2. Improves Model Performance:
es
○ Selecting the most relevant features can enhance the
model’s accuracy and predictive power. It allows the model
ot
to focus on the most informative data points, which can lead
to better generalization to unseen data.
3. Enhances Interpretability:
N
○ A model with fewer features is often easier to interpret
and understand. This is especially important in fields such as
's
healthcare and finance, where stakeholders need to
understand the factors driving predictions.
ah
4. Reduces Computational Cost:
○ Fewer features lead to a simpler model that requires less
hm
computational resources, which is particularly beneficial for
large datasets. It speeds up the training process and
reduces the time and memory required for both training and
prediction.
Re
5. Addresses the Curse of Dimensionality:
○ In high-dimensional spaces, the amount of data required to
make reliable predictions increases exponentially. Feature
selection mitigates this issue by reducing the
dimensionality, allowing the model to perform better with
limited data.
6. Improves Data Quality:
○ Feature selection can help identify and remove noisy or
irrelevant features that do not contribute meaningfully to
the analysis, leading to higher-quality datasets and better
model performance.
7. Facilitates Model Selection:
○ Different machine learning algorithms may require
es
different features for optimal performance. Feature
selection can help identify the most relevant features for
ot
each algorithm, aiding in model comparison and selection.
Usefulness in Machine Learning Algorithms
1. Enhanced Learning:
N
○ Machine learning algorithms perform better when they
's
focus on the most relevant features. Feature selection
helps in identifying those features that contribute most
ah
significantly to the output, leading to improved learning.
2. Faster Training:
○ Training times are reduced when fewer features are used.
hm
This is especially important for algorithms like Support
Vector Machines (SVM), Random Forest, or Neural
Networks, which can be computationally intensive.
Re
3. Better Generalization:
○ Models built with relevant features are more likely to
generalize well to new, unseen data. This leads to better
performance in real-world applications where the model
encounters data it has not seen before.
4. Support for Specific Algorithms:
○ Some algorithms, like decision trees, can benefit
significantly from feature selection. By reducing the number
of features, decision trees can create simpler models that
make better splits and predictions.
5. Increased Robustness:
○ Feature selection can lead to models that are less sensitive
to variations in the data, making them more robust in the
es
presence of noise or outliers.
Methods of Feature Selection
ot
Feature selection can be performed using various methods, including:
N
● Filter Methods: Evaluate features based on statistical measures
(e.g., correlation, Chi-square test) to select relevant features
independent of the learning algorithm.
's
ah
● Wrapper Methods: Use a specific machine learning algorithm to
hm
evaluate combinations of features and select the best-performing
subset.
Re
● Embedded Methods: Perform feature selection during the model
training process (e.g., Lasso regression, which adds a penalty for
including too many features).
Q18) State and explain common errors in machine learning / 6
es
Common Mistakes Machine Learning
ot
5 Common Machine Learning Errors
● Lack of understanding the mathematical aspect of machine
learning algorithms
N
● Data Preparation and Sampling
○ Data Cleansing
's
○ Feature Engineering
○ Sampling
ah
● Implementing machine learning algorithms without a strategy
● Implementing everything from scratch
● Ignoring outliers
hm
References
Re
Research Paper on Cognitive automation by Christian Engel1 · Philipp
Ebel1 · Jan Marco Leimeister1
https://www.javatpoint.com/
https://www.geeksforgeeks.org/
https://www.kaggle.com/
You might also like
- Machine Learning PPT For StudentsDocument18 pagesMachine Learning PPT For Studentsrajesh78% (9)
- Great Joint International 10b 8feb FinalDocument5 pagesGreat Joint International 10b 8feb Finalhamid nabizade92% (12)
- Machine Learning NotesDocument19 pagesMachine Learning NotesAbir Baidya100% (7)
- Introduction To Machine LearningDocument42 pagesIntroduction To Machine Learning22ht1a05b9No ratings yet
- Unit 1Document32 pagesUnit 1vamsi kiranNo ratings yet
- Chapter 2. Literature SurveyDocument7 pagesChapter 2. Literature SurveyJai Manohar BandreNo ratings yet
- Machine Learning 3Document30 pagesMachine Learning 3suvajit2021No ratings yet
- Machine Learning Unit 4Document78 pagesMachine Learning Unit 4Niki GowthuNo ratings yet
- T Thesis Topics in Machine Learning For Research ScholarsDocument14 pagesT Thesis Topics in Machine Learning For Research ScholarsmuraliNo ratings yet
- Introduction To Machine LearningDocument78 pagesIntroduction To Machine LearningOmkar MasayeNo ratings yet
- Module 4Document28 pagesModule 4sushma-icbNo ratings yet
- ML Unit 1Document20 pagesML Unit 1shiva751514No ratings yet
- Null 5Document16 pagesNull 5eshakalifathimaNo ratings yet
- DAIOT UNIT 5 (1) OwnDocument13 pagesDAIOT UNIT 5 (1) OwnchintuNo ratings yet
- Unit 1Document43 pagesUnit 1amrutamhetre9No ratings yet
- Unit 1Document95 pagesUnit 1canikissyou7984No ratings yet
- ML 1Document79 pagesML 1midnight.emprorerNo ratings yet
- Review Paper On Machine LearningDocument4 pagesReview Paper On Machine LearningShubham Sah100% (1)
- TIS - Intro To Machine LearningDocument18 pagesTIS - Intro To Machine Learningapoorva.k2017No ratings yet
- Unit 1 ml2Document23 pagesUnit 1 ml2Srujana ShettyNo ratings yet
- Machine Learning 1Document34 pagesMachine Learning 1suvajit2021No ratings yet
- Unit - 3 - MLDocument53 pagesUnit - 3 - MLaakhi0887No ratings yet
- ML 1Document35 pagesML 1atuldev8132No ratings yet
- DSF - UNIT III NotesDocument17 pagesDSF - UNIT III NotesRockerz RickNo ratings yet
- Unit - 1 Machine LearningDocument82 pagesUnit - 1 Machine LearningAbhishekNo ratings yet
- COSC 210 INTRODUCTION TO MACHINE LEARNING Module I-1Document46 pagesCOSC 210 INTRODUCTION TO MACHINE LEARNING Module I-1Ustäz Däñ MätërwälléNo ratings yet
- Unit 1 Machine LearningDocument26 pagesUnit 1 Machine LearningAviral GuptaNo ratings yet
- ML Unit 1Document25 pagesML Unit 1Sristi Jaya MangalaNo ratings yet
- AI Session 3 Machine Learning SlidesDocument35 pagesAI Session 3 Machine Learning SlidesPhilani MangeziNo ratings yet
- Unit 1Document8 pagesUnit 1vvvcxzzz3754No ratings yet
- Machine LearningDocument103 pagesMachine Learningsp1135220No ratings yet
- Machine Learning Unit 1Document72 pagesMachine Learning Unit 1Manshi JainNo ratings yet
- UNIT 1 All NotesDocument24 pagesUNIT 1 All Notesniku13102001No ratings yet
- Machine Learning1Document11 pagesMachine Learning1Rishab Bhattacharyya100% (1)
- Machine LearningDocument12 pagesMachine Learning20CM052-Shravani Shinganwadikar. CMNo ratings yet
- Machine LearningDocument13 pagesMachine LearningAini RizviNo ratings yet
- Deep Learning FileDocument60 pagesDeep Learning FileAyush singhNo ratings yet
- Unit 2Document63 pagesUnit 2Mdv PrasadNo ratings yet
- Machine Learning Question Bank - (NOT GIVEN BY MAM)Document50 pagesMachine Learning Question Bank - (NOT GIVEN BY MAM)Anushka ManeNo ratings yet
- Introduction To Machine LearningDocument4 pagesIntroduction To Machine LearningSHUBHAM SRIVASTAVANo ratings yet
- Flow Diagram of Machine Learning or Life Cycle of Machine LearningDocument91 pagesFlow Diagram of Machine Learning or Life Cycle of Machine LearningJay MangukiyaNo ratings yet
- Chapter-4 - Intro To Machine LearningDocument26 pagesChapter-4 - Intro To Machine LearningsindhuNo ratings yet
- Everything You Need To Know About Machine LearningDocument6 pagesEverything You Need To Know About Machine LearningPrn BstNo ratings yet
- MLT Q and ADocument19 pagesMLT Q and Aharigoutham627No ratings yet
- ML FinalDocument28 pagesML FinalShreya100% (1)
- My ML NotesDocument6 pagesMy ML NotesPandu KNo ratings yet
- Machine Learning IntroDocument27 pagesMachine Learning IntroChandana NarraNo ratings yet
- Machine Learning: Bilal KhanDocument26 pagesMachine Learning: Bilal KhanBilal KhanNo ratings yet
- Module 1 PPTDocument122 pagesModule 1 PPTharkamals.it.21No ratings yet
- Machine LearningDocument11 pagesMachine LearningTonpai DolnapaNo ratings yet
- Deep Learning FileDocument58 pagesDeep Learning FileAyush singhNo ratings yet
- Part 2 Introduction To MLDocument13 pagesPart 2 Introduction To MLsujatakayaNo ratings yet
- Unit 1 NotesDocument29 pagesUnit 1 NotessakthiasphaltalphaNo ratings yet
- ML Notes Unit 1-2Document55 pagesML Notes Unit 1-2Mudit MattaNo ratings yet
- Unit 1 - Machine LearningDocument17 pagesUnit 1 - Machine LearningASHOKA KUMARNo ratings yet
- Mechine LearningDocument106 pagesMechine Learningkrritik3210No ratings yet
- An Overview of Machine LearningDocument20 pagesAn Overview of Machine Learningsomeya mohsenNo ratings yet
- ML - Module 1Document30 pagesML - Module 1Amritesh KumarNo ratings yet
- Reinforcement Learning: Parallelizing Genetic AlgorithmsDocument5 pagesReinforcement Learning: Parallelizing Genetic AlgorithmsDHARINI R SIT 2020No ratings yet
- Big-Data Unit-3Document54 pagesBig-Data Unit-3Tulshiram Kamble100% (1)
- Key Concepts Example of A Control ChartDocument2 pagesKey Concepts Example of A Control ChartBoran YelkenciogluNo ratings yet
- Image Enhancement-Spatial Filtering From: Digital Image Processing, Chapter 3Document56 pagesImage Enhancement-Spatial Filtering From: Digital Image Processing, Chapter 3Jaunty_UNo ratings yet
- ASCE710W - ASCE 7-10 Code Wind Analysis ProgramDocument2 pagesASCE710W - ASCE 7-10 Code Wind Analysis Programmr.Kram0% (1)
- ABB DCS Life - Cycle - Parts - Services by FamilyDocument16 pagesABB DCS Life - Cycle - Parts - Services by FamilyArdvarkNo ratings yet
- HW 2 Chap 1 Jxip Final PDFDocument11 pagesHW 2 Chap 1 Jxip Final PDFJuan Xavier Izurieta Pineda100% (1)
- 135+ Pages-Modern India History Class Notes For UPSC Civil Service Prelims, APPSC & SSCDocument11 pages135+ Pages-Modern India History Class Notes For UPSC Civil Service Prelims, APPSC & SSCAshok PediredlaNo ratings yet
- AIS AllSlidesDocument92 pagesAIS AllSlidesMario ZaraNo ratings yet
- Patient CentreDocument2 pagesPatient CentredesbestNo ratings yet
- BC547Document4 pagesBC547zigobasNo ratings yet
- Bluetti EB70 ManualDocument28 pagesBluetti EB70 Manualcheff96No ratings yet
- SolutionsDocument9 pagesSolutionsdavid_benavidez_3No ratings yet
- The Nexus Integration TeamDocument5 pagesThe Nexus Integration TeamAlberto Caetano100% (1)
- Unit - IiDocument45 pagesUnit - IiJit AggNo ratings yet
- FMNF05 Assignment3Document21 pagesFMNF05 Assignment3Bengt HörbergNo ratings yet
- What Is A Function?: Advantage of PHP FunctionsDocument12 pagesWhat Is A Function?: Advantage of PHP FunctionsSIDNY BASCONo ratings yet
- Cybersecurity Framework Profile Ransomware Risk MG Beckman 1623394907Document22 pagesCybersecurity Framework Profile Ransomware Risk MG Beckman 1623394907Chandrashekar MNo ratings yet
- Technology System Support Senior in Baltimore MD Resume George "Terry" WhiteDocument1 pageTechnology System Support Senior in Baltimore MD Resume George "Terry" WhiteGeorgeTerryWhiteNo ratings yet
- CLL F394 TRM EngDocument10 pagesCLL F394 TRM EngMarcio MarquesNo ratings yet
- Motor StartersDocument4 pagesMotor Starterssui1981No ratings yet
- IBM Software Business Partner Guide: September 2012Document126 pagesIBM Software Business Partner Guide: September 2012Eric Andrianto100% (1)
- Malik Et Al. - 2023Document26 pagesMalik Et Al. - 2023apatzNo ratings yet
- Pi20221228 VN 3Document1 pagePi20221228 VN 3pcatruongNo ratings yet
- Ansoft Solutions Filter 8029Document8 pagesAnsoft Solutions Filter 8029MohammadYanuarNo ratings yet
- STAT 544.01 - Topics - Spatial StatisticsDocument4 pagesSTAT 544.01 - Topics - Spatial StatisticswerwerwerNo ratings yet
- BlackBox Related CommandsDocument3 pagesBlackBox Related Commandsshantibhushan.saleNo ratings yet
- CV - Shafiqullah AhmadzaiDocument3 pagesCV - Shafiqullah AhmadzaiShafiq AhmadzaiNo ratings yet
- Theory - ComputationDocument97 pagesTheory - ComputationUmamaheswar PutrevuNo ratings yet
- Ict Css 12 q1 Las1 FinalDocument8 pagesIct Css 12 q1 Las1 FinalPaul Senen DiduloNo ratings yet
- Social MEdia and Self - Influence On The Formation of Identity and Understanding of SelfDocument46 pagesSocial MEdia and Self - Influence On The Formation of Identity and Understanding of SelfArninz BernatNo ratings yet
