UNIT 1

Q1. How can the Company use Hadoop and Map Reduce. Explain Process and key benefits of these tools.
(Case Study).
Ans- 1. Introduction to Hadoop and MapReduce:
 Hadoop is an open-source framework for distributed storage and processing of large data sets.
MapReduce is a programming model for processing data in parallel on a cluster.
2. Data Storage with HDFS:
 HDFS stores data across multiple nodes, ensuring high availability and fault tolerance by
replicating data blocks.
3. Data Processing with MapReduce:
 MapReduce processes data in two steps: Map filters and sorts data, while Reduce summarizes results,
enabling efficient parallel processing.
4. Handling Big Data:
 Hadoop and MapReduce handle vast amounts of structured and unstructured data, suitable for
industries like finance, healthcare, and retail.
5. Scalability:
 Hadoop clusters can easily scale by adding more nodes, allowing cost-effective handling of growing
data volumes.
6. Cost-Effectiveness:
 Hadoop is open-source, reducing the need for expensive proprietary solutions, and can run on
commodity hardware.
7. Fault Tolerance and Reliability:
 Hadoop manages data replication and distributes data across nodes, ensuring fault tolerance and high
reliability during node failures.
8. Flexibility and Versatility:
 Hadoop processes various types of data from multiple sources, and MapReduce allows custom data
processing algorithms to meet specific business needs.
Q2. Role of Data Scientists. What skills and tools are used for fraud detection Model. (Case Study)
Ans- 1. Data Collection and Preprocessing:
 Data scientists gather data from various sources and clean it to remove inaccuracies. This ensures
high-quality data for analysis and modeling.
2. Exploratory Data Analysis (EDA):
 EDA involves analyzing data sets to summarize their main characteristics, often using visual
methods. This helps identify patterns, trends, and anomalies crucial for fraud detection.
3. Feature Engineering:
 Creating and selecting relevant features from raw data is essential. Data scientists use domain
knowledge to craft features that improve model performance.
4. Model Development:
 Building and training machine learning models such as logistic regression, decision trees, and
neural networks tailored to detect fraudulent activities.
 5. Model Evaluation and Validation:
 Evaluating models using metrics like precision, recall, F1 score, and ROC-AUC ensures they
effectively distinguish between fraudulent and legitimate activities.
6. Deployment and Monitoring:
 After validation, models are deployed in real-time systems. Continuous monitoring ensures
models remain effective as fraud patterns evolve.
7. Collaboration with Stakeholders:
 Data scientists work with business analysts, IT teams, and other stakeholders to integrate fraud
detection models into the company's operational systems.
8. Continuous Learning and Improvement:
 Keeping abreast of the latest developments in data science and fraud detection to continuously
improve models and methodologies.
9. Programming Skills:
 Proficiency in languages like Python and R is essential for data manipulation, statistical analysis,
and implementing machine learning algorithms.
10. Tools and Technologies:
 Utilizing tools like Pandas and NumPy for data manipulation, Matplotlib and Seaborn for data
visualization, and Scikit-learn, TensorFlow, and Keras for building machine learning models.
11. Big Data Technologies:
 Knowledge of Hadoop, Spark, and Hive is useful for processing large datasets, making data
scientists more efficient in handling big data for fraud detection.
12. Database Management:
 Skills in SQL and NoSQL databases enable data scientists to efficiently query and manage data
from different sources, ensuring comprehensive analysis.
Q3. Specific Features of Madlib and insights gain impact decision making and risk management. (Case
Study)
Ans- 1. Scalability and Parallel Processing:
 MADlib uses MPP databases like Greenplum, enabling efficient handling of large-scale data
analytics for timely insights.
2. Advanced Analytical Functions:
 Offers a range of statistical and machine learning functions for building predictive models to
forecast risks and opportunities.
3. Integration with SQL:
 Operates within SQL databases, allowing analysts to perform complex analytics using familiar
SQL queries, simplifying workflows.
4. Open Source and Extensibility:
 Continuously updated by the developer community, ensuring access to the latest techniques and
customization for specific needs.
 5. Data Preprocessing Capabilities:
 Includes tools for data transformation and handling missing values, crucial for accurate model
building and reliable risk assessment.
6. Predictive Modeling and Forecasting:
 Enables identification of potential risks and opportunities by analyzing historical data, supporting
proactive decision-making.
7. Real-time Analytics:
 Processes and analyzes data in real time, allowing organizations to detect anomalies and manage
risks as they occur.
8. Cost Efficiency:
 Integrates advanced analytics within existing SQL databases, reducing infrastructure costs and
enhancing resource allocation for risk management.
Q4. Design Data Analytics Framework, Data source and Analytic method.
Ans- 1. Data Collection:
 Gather data from various sources such as databases, APIs, web scraping, and IoT devices. Ensure
the data is relevant, accurate, and comprehensive for the analysis.
2. Data Ingestion:
 Use ETL (Extract, Transform, Load) processes to move data into a centralized storage system like
a data warehouse or data lake. Tools like Apache NiFi or Talend can streamline this process.
3. Data Storage:
 Store the ingested data in scalable storage solutions like Hadoop HDFS, Amazon S3, or a
relational database (e.g., PostgreSQL). Ensure the storage system supports the volume and variety
of data collected.
4. Data Preprocessing:
 Clean, normalize, and transform the data to prepare it for analysis. Use tools like Apache Spark or
Pandas for handling missing values, outliers, and standardizing formats.
5. Data Exploration and Visualization:
 Perform exploratory data analysis (EDA) to understand data patterns and relationships. Utilize
visualization tools like Tableau, Power BI, or Matplotlib to create insightful visual representations.
6. Analytic Methods:
 Apply statistical analysis, machine learning algorithms (e.g., regression, classification, clustering),
and predictive modeling. Use frameworks like Scikit-learn, TensorFlow, or R for developing
models.
7. Model Deployment:
 Deploy the developed models into production environments using platforms like Docker,
Kubernetes, or cloud services (e.g., AWS SageMaker). Ensure models can handle real-time data
and scale as needed.
8. Continuous Monitoring and Maintenance:
 Monitor model performance and accuracy over time. Use tools like MLflow or Prometheus to
track metrics, and continuously update models based on new data and changing patterns.
 UNIT-2
Q.1 How company redesign its analytical structure and explain suitable Big Data Architecture that’s
improve elements data ingestion, storage, processing and analysis.
Ans- 1. Data Ingestion Layer:
 Implement scalable ingestion tools like Apache Kafka or Flume to efficiently collect data from
various sources in real-time.
2. Data Storage Layer:
 Use distributed storage solutions like Hadoop HDFS or Amazon S3 to store large volumes of
structured and unstructured data.
3. Data Processing Layer:
 Adopt Apache Spark or Flink for fast, in-memory data processing, supporting both batch and real-
time analytics.
4. Data Integration:
 Utilize ETL tools like Apache Nifi or Talend to streamline data transformation and integration
across different systems.
5. Data Analytics Platform:
 Integrate analytics platforms like Apache Hive or Presto to enable SQL-based querying and
analysis of large datasets.
6. Machine Learning and AI:
 Leverage machine learning frameworks such as TensorFlow or Scikit-learn for building and
deploying predictive models.
7. Data Visualization:
 Use visualization tools like Tableau or Power BI to create interactive dashboards and reports for
insights and decision-making.
8. Data Governance and Security:
 Implement robust data governance frameworks and security measures, ensuring compliance and
protecting sensitive information.
Q2. Design Big data Architecture framework.
Ans-
1. Data Sources Layer:
 Components: Structured data (databases), semi-structured data (logs, XML/JSON files), unstructured
data (videos, images, social media).
 Function: Collect data from various internal and external sources.
2. Data Ingestion Layer:
 Components: Apache Kafka, Apache Flume, AWS Kinesis.
 Function: Capture and load real-time and batch data into the system efficiently.
3. Data Storage Layer:
  Components: Hadoop HDFS, Amazon S3, Google Cloud Storage.
 Function: Store large volumes of structured, semi-structured, and unstructured data.
4. Data Processing Layer:
 Components: Apache Spark, Apache Flink, Apache Storm.
 Function: Perform batch and real-time data processing and transformation.
5. Data Integration and ETL Layer:
 Components: Apache Nifi, Talend, Informatica.
 Function: Extract, transform, and load data into appropriate storage and processing systems.
6. Data Analytics Layer:
 Components: Apache Hive, Apache Impala, Presto.
 Function: Provide SQL-based querying and analysis capabilities on large datasets.
7. Machine Learning and Advanced Analytics Layer:
 Components: TensorFlow, Scikit-learn, Spark MLlib.
 Function: Develop, train, and deploy machine learning models for predictive and prescriptive
analytics.
8. Data Visualization and BI Layer:
 Components: Tableau, Power BI, Apache Superset.
 Function: Create interactive dashboards and visualizations for business intelligence and decision-
making.
9. Data Governance and Security Layer:
 Components: Apache Ranger, Apache Atlas, data encryption tools.
 Function: Ensure data quality, security, compliance, and governance across the data lifecycle.
10. Monitoring and Management Layer:
 Components: Prometheus, Grafana, Nagios.
 Function: Monitor system performance, resource utilization, and ensure the health of the big data
environment.
Q3. Analyze Current Analytical Architecture. How they work together to enable real time analytics.
Ans-
1. Data Sources Layer:

 Data from multiple sources, such as IoT devices, social media feeds, and databases, is
continuously collected. This diverse data forms the foundation for real-time analytics, ensuring a
broad range of insights.

2. Data Ingestion Layer:

 Real-time data streaming tools like Apache Kafka or AWS Kinesis capture and transport data.
These tools enable the ingestion of high-volume data streams without delay, ensuring data is ready
for immediate processing.

3. Data Storage Layer:

  Real-time data is stored in fast-access storage systems like HDFS, NoSQL databases (Cassandra,
MongoDB), or in-memory databases. This layer ensures that incoming data is available for quick
retrieval and analysis.

4. Data Processing Layer:

 Frameworks like Apache Spark Streaming and Apache Flink process data in real-time by applying
computations, transformations, and aggregations. This allows for near-instant insights and helps
organizations act quickly on data.

5. Data Integration and ETL Layer:

 Data integration tools such as Apache Nifi or Talend ensure seamless extraction, transformation,
and loading (ETL) of data. This layer maintains data consistency and quality, ensuring it is ready
for accurate analysis.

6. Data Analytics Layer:

 SQL-based analytic tools like Presto or Apache Hive enable users to run fast queries on large
datasets. These tools support low-latency analytics, providing real-time answers to business
questions.

7. Machine Learning and Advanced Analytics Layer:

 Real-time data feeds into machine learning models built using tools like TensorFlow or Spark
MLlib. These models can instantly predict trends, detect anomalies, or recommend actions based
on incoming data.

8. Data Visualization and BI Layer:

 Visualization tools such as Tableau or Grafana transform real-time data insights into actionable
visual dashboards. These dashboards update in real-time, enabling businesses to make data-driven
decisions immediately.

9. Monitoring and Security Layer:

 Continuous monitoring tools like Prometheus and Grafana ensure that the system is operating
smoothly. Data governance and security tools protect the integrity and privacy of data, ensuring
compliance and system stability.
 UNIT 3

Q1. How would you apply the data into Analytics Lifecycle. Describe main phases and specific roles and
responsibilities of member in each stage.
Ans- To apply data into the Analytics Lifecycle, follow these main phases along with the specific roles and
responsibilities of team members in each stage:
1. Discovery:
 Roles: Business analysts, domain experts, and project managers.
 Responsibilities: Define the business problem, set objectives, and identify data sources and
stakeholders involved.
2. Data Preparation:
 Roles: Data engineers, data scientists, and database administrators.
 Responsibilities: Collect, clean, and transform raw data into a usable format, ensuring data
quality and consistency.
3. Model Planning:
 Roles: Data scientists and statisticians.
 Responsibilities: Choose appropriate modeling techniques, define algorithms, and establish a
clear plan for model development.
4. Model Building:
 Roles: Data scientists and machine learning engineers.
 Responsibilities: Develop and train predictive models using prepared data, iterating to
improve accuracy and performance.
5. Evaluation:
 Roles: Data scientists and business analysts.
 Responsibilities: Assess model performance using metrics and validation techniques,
ensuring it meets business requirements and objectives.
6. Deployment:
 Roles: Data engineers, software developers, and IT support.
 Responsibilities: Implement the model into production systems, integrating it with existing
processes and ensuring it operates smoothly.
7. Monitoring and Maintenance:
 Roles: Data scientists, IT support, and operations teams.
 Responsibilities: Continuously track model performance, update it as necessary, and address
any issues or changes in the data environment.
 8. Communicating Results:
 Roles: Business analysts, project managers, and data visualization experts.
 Responsibilities: Present findings and insights to stakeholders through reports, dashboards,
and presentations, ensuring clarity and actionable recommendations.

Q2. Design an outline for stakeholder at different stages.

Ans-

1. Discovery:

 Stakeholders: Business analysts, domain experts, project managers, senior management.

 Outline: Define business objectives, identify key questions, and establish the project scope
and timeline. Engage stakeholders to understand requirements and expectations.

2. Data Preparation:

 Stakeholders: Data engineers, data scientists, database administrators.

 Outline: Gather relevant data sources, clean and transform the data, and ensure data quality.
Document data processes and create a data dictionary.

3. Model Planning:

 Stakeholders: Data scientists, statisticians, business analysts.

 Outline: Select modeling techniques and algorithms, define model evaluation criteria, and
develop a project plan. Conduct exploratory data analysis to understand data patterns.

4. Model Building:

 Stakeholders: Data scientists, machine learning engineers, software developers.

 Outline: Develop and train predictive models, optimize parameters for performance, and
validate models. Document the model development process for transparency.

5. Evaluation:

 Stakeholders: Data scientists, business analysts, domain experts.

 Outline: Assess model performance using defined criteria, perform cross-validation, and
interpret results in a business context. Gather feedback from stakeholders for refinement.

6. Deployment:

 Stakeholders: Data engineers, software developers, IT support, operations teams.

 Outline: Integrate the model into production systems, ensuring scalability and reliability.
Establish monitoring procedures and train end-users on model usage.

7. Monitoring and Maintenance:

 Stakeholders: Data scientists, IT support, operations teams.

 Outline: Continuously monitor model performance and update it with new data. Address
issues or anomalies and maintain documentation and version control.

8. Communicating Results:
  Stakeholders: Business analysts, project managers, data visualization experts, senior
management.
 Outline: Present findings and insights using reports and dashboards. Provide actionable
recommendations and gather feedback for future improvements.

Q3. Discuss the phases of Data Analytics Lifecycle.

Ans- 1. Discovery:

 Description: The initial phase focuses on understanding the business problem and
defining project objectives. Stakeholders are identified, and the project scope and timeline
are established.
 Key Activities: Engaging with stakeholders, gathering requirements, and defining
success criteria.

2. Data Preparation:

 Description: This phase involves collecting, cleaning, and transforming raw data into a
usable format. It ensures the data's quality and readiness for analysis.
 Key Activities: Data extraction, data cleaning, dealing with missing values, and data
transformation.

3. Model Planning:

 Description: During this phase, analysts and data scientists select appropriate modeling
techniques and algorithms. A plan for model development is outlined based on the data's
characteristics.
 Key Activities: Exploratory data analysis, feature selection, and deciding on modeling
approaches.

4. Model Building:

 Description: In this phase, predictive models are developed and trained using the
prepared data. Iterative testing and refinement are conducted to optimize model
performance.
 Key Activities: Model development, training, hyperparameter tuning, and validation.

5. Evaluation:

 Description: The developed models are rigorously evaluated to ensure they meet the
business objectives and perform well on validation data. This phase involves assessing
model accuracy and reliability.
 Key Activities: Performance metrics calculation, cross-validation, and model
interpretation.

6. Deployment:

 Description: Successful models are deployed into production systems where they can be
used for making real-time decisions. This phase ensures that the model integrates
seamlessly with existing workflows.
 Key Activities: Model integration, implementation, monitoring, and user training.
 7. Monitoring and Maintenance:

 Description: After deployment, models are continuously monitored to maintain their

performance. Necessary updates and adjustments are made in response to changes in data
or business requirements.
 Key Activities: Performance tracking, model updating, and addressing any operational
issues.

8. Communicating Results:

 Description: Insights and results from the analysis are communicated to stakeholders
through reports and visualizations. This phase ensures that the findings are understood
and can inform decision-making.
 Key Activities: Creating dashboards, preparing presentations, and providing actionable
recommendations.

Q4. Identify the key roles for stakeholder through analytics project.

Ans-

1. Business Sponsor:

 Responsibilities: Provides overall direction and funding for the project. Ensures the project
aligns with strategic business goals and priorities.

2. Project Manager:

 Responsibilities: Manages project timelines, resources, and communication among

stakeholders. Ensures the project stays on track and meets deadlines.

3. Business Analyst:

 Responsibilities: Acts as a bridge between business stakeholders and the technical team.
Defines business requirements, objectives, and key performance indicators (KPIs).

4. Data Engineer:

 Responsibilities: Prepares the data infrastructure, including data collection, cleaning, and
transformation. Ensures data quality and accessibility for analysis.

5. Data Scientist:

 Responsibilities: Develops and implements analytical models and algorithms. Conducts

exploratory data analysis, model training, and validation.

6. Statistical Analyst:

 Responsibilities: Applies statistical methods to analyze data and interpret results. Helps in
selecting appropriate statistical techniques and validating models.

7. Domain Expert:
  Responsibilities: Provides domain-specific knowledge and context to the data analysis.
Ensures the analysis considers industry-specific factors and insights.

8. IT Support:

 Responsibilities: Provides technical support for data storage, processing, and deployment.
Ensures the infrastructure is secure, scalable, and reliable.

9. End Users:

 Responsibilities: Utilize the analytical tools and insights generated by the project. Provide
feedback on usability and effectiveness of the solutions.

10. Data Governance Officer:

 Responsibilities: Ensures data privacy, security, and compliance with regulations. Develops
and enforces data governance policies.

11. Visualization Expert:

 Responsibilities: Creates intuitive and effective data visualizations and dashboards. Helps in
communicating complex insights to non-technical stakeholders.

12. Change Management Specialist:

 Responsibilities: Manages the impact of project outcomes on business processes. Ensures

smooth transition and adoption of new tools and insights.
 UNIT 4 (2M)
Q1. Two Benefits of ‘R’.
Ans-
1. Extensive Statistical and Data Analysis Capabilities:
 R provides a wide range of statistical techniques and data analysis tools, including linear
and nonlinear modeling, time-series analysis, and clustering. Its comprehensive
functionality makes it ideal for conducting detailed and complex analyses.
2. Strong Community and Vast Package Ecosystem:
 R has a large and active community that continuously contributes to its extensive package
ecosystem. Thousands of packages are available through CRAN, offering tools for data
manipulation, visualization, machine learning, and more, making it highly versatile for
various analytical tasks.

Q2. Data Frame in ‘R’.

Ans-
Data frame in R is a versatile data structure used to store tabular data. It consists of rows and
columns, where each column can contain different types of data (numeric, character, factor, etc.).
Data frames are ideal for handling datasets in R because they allow for easy manipulation, analysis,
and visualization of data. They can be created using functions like [Link]() and can be easily
subsetted, merged, and transformed using various R functions.

Q3. Difference Between Vector and Lists in ‘R’.

Ans-
Feature Vector List
Homogeneous: all elements must be Heterogeneous: elements can be of
Data Type
of the same type different types
Element
Single bracket [ ] notation Double bracket [[ ]] notation
Access
Simple, single, continuous block of Complex, can contain different types
Structure
memory and even other lists
Suitable for simple data types and Suitable for storing complex and mixed
Use Case
arithmetic operations data types
list(1, "a", TRUE, c(1, 2, 3)) (mixed
Examples c(1, 2, 3, 4) (all numeric)
types)

Q4. Calculate mean in ‘R’.

Ans-
1. Mean
 It is the sum of observation divided by the total number of observations. It is also
defined as average which is the sum divided by count.
2. What is Mean?
  The mean is the average of a set of numbers. It is calculated by summing all the
values in the dataset and dividing by the total number of values.
# R program to illustrate
# Descriptive Analysis
# Import the data using [Link]()
myData = [Link]("[Link]", stringsAsFactors = F)
# Compute the mean value
mean = mean(myData$Age)
Q5. Basic R-Script to load and plain datasets.
Ans-
# Load a dataset (e.g., a CSV file)
data <- [Link]("path/to/your/[Link]")

# View the first few rows of the dataset

head(data)

# Display summary statistics of the dataset

summary(data)

# Check the structure of the dataset

str(data)

Q6. Calculate Mean in ‘R’ command.

Ans- (Refer Q4 above)

Q7. Difference between Co-relation and Co-variance in ‘R’.

Ans-

Q8. Given two linear regression Model, which model you will prefer.
Ans-
1. R-squared (R²) Value:
 Prefer the model with the higher R² value, as it indicates how well the model explains
the variability of the dependent variable. A higher R² suggests a better fit to the data.
2. Mean Squared Error (MSE) or Root Mean Squared Error (RMSE):
 Choose the model with the lower MSE or RMSE, as these metrics measure the
average squared difference between observed and predicted values. Lower values
indicate better prediction accuracy.
3. Adjusted R-Squared:
If both models have different numbers of predictors, we should prefer the model with
the higher Adjusted R², which accounts for both the fit and complexity.
Preferred model: The one with the higher Adjusted R².

Q9. Two popular libraries in ‘R’.

Ans-
1. ggplot2:
 Description: A powerful library for data visualization in R. It provides a flexible and
consistent way to create a variety of static, interactive, and animated plots based on
the Grammar of Graphics.
 Use: Used for creating complex plots like bar charts, line graphs, histograms, and
scatter plots with ease.
2. dplyr:
 Description: A part of the tidyverse, dplyr is a library for data manipulation. It
provides a set of functions to filter, select, mutate, arrange, and summarize data
efficiently.
 Use: Used for performing common data manipulation tasks such as subsetting,
sorting, grouping, and summarizing large datasets.

Q10. Difference between Vector and Lists in ‘R’.

Ans-

(6 MARKS)

Q1. Using ‘R’ determine which variable have most significant impact on House price in detail.
Ans-
Steps to Determine Significant Variables Impacting House Price:
1. Load and Explore the Dataset:
 Begin by loading the dataset containing house prices and potential predictors
(e.g., square footage, number of bedrooms, location, etc.). View the data to
understand its structure and contents.
 2. Data Preprocessing:
 Clean the dataset by checking for missing values and handling them
appropriately (either by imputation or removal). Ensure all variables are in the
correct format (e.g., numeric or categorical).
3. Fit a Linear Regression Model:
 Use the lm() function to fit a multiple linear regression model with house price
as the dependent variable and other variables as independent predictors.

4. Interpret the Model Summary:

 Use the summary() function to view the model’s output. Key statistics to focus
on:
 Coefficients: These represent the estimated impact of each variable on
house price.
 p-values: A p-value less than 0.05 indicates that a variable is
statistically significant.
 R-squared: Shows how well the model explains the variance in house
prices.
5. Identify the Most Significant Variable:
 Look at the p-values associated with each predictor variable. The variable with
the smallest p-value (typically less than 0.05) is the most statistically
significant in affecting house prices.
 Consider the coefficients: Higher absolute values of the coefficients indicate a
larger impact on the house price.
6. Conclusion:
 The variable with the lowest p-value and a high absolute coefficient (if
positive) has the most significant impact on house price. Other variables may
also show significance but with a lesser effect.

Q2. Difference Between Overfitting and Underfitting, How ‘R’ is used to detect this issue.
Ans-
1. Cross-Validation:
 Cross-validation is used to assess the model's performance by splitting the data into subsets
(folds) and training the model multiple times. It helps in detecting overfitting when the model
performs well on training data but poorly on validation data, and underfitting when the model
fails to perform well on both training and validation sets.
2. Learning Curves:
 Learning curves plot the training and test errors as the size of the training set increases.
Overfitting is indicated when the training error continues to decrease while the test error
increases, and underfitting is indicated by both errors being high and not improving with more
data.
3. Train-Test Split:
 By splitting the data into training and test sets, you can evaluate the model's performance on
unseen data. If the model performs well on the training set but poorly on the test set, it
indicates overfitting. Conversely, poor performance on both sets suggests underfitting.
4. Model Evaluation Metrics:
 Metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-
squared help assess the model's accuracy. High performance on the training set but poor
performance on the test set signals overfitting. Underfitting is evident when both training and
test performance are poor.
5. Regularization Techniques:
 Regularization methods like Lasso and Ridge regression are used to reduce overfitting by
penalizing overly complex models. These techniques help prevent the model from learning
noise in the data and promote simpler, more generalizable models.

Q3. How these techniques summarization, visualization help in building statistical method,
Example of functions in R for these purpose.
Ans-
1. Summarization:
 summary(): Provides a summary of the dataset, including measures like mean,
median, minimum, and maximum values.
  mean(), median(), sd(): Calculate the mean, median, and standard deviation for
numerical columns.
 table(): Creates frequency tables for categorical variables, helping to summarize
counts of each category.
 cor(): Computes the correlation matrix between numerical variables, helping to
understand relationships between variables.
2. Visualization:
 Base R:
 plot(): Used for creating scatter plots to visualize relationships between two
continuous variables.
 hist(): Creates histograms to visualize the distribution of a single variable.
 boxplot(): Displays box plots to identify the spread and outliers in the data.
 ggplot2 (Advanced Visualization):
 ggplot(): Used for creating complex plots with layers for various types of
data (e.g., bar charts, line plots, histograms).
 geom_point(), geom_histogram(), geom_boxplot(): These functions within
ggplot2 help create specific plot types like scatter plots, histograms, and box
plots, respectively.

Summary:
 Summarization helps in understanding the dataset by reducing complex data into simpler
statistics, ensuring the data is clean, and identifying patterns before applying statistical
models.
 Visualization provides a visual understanding of the data, making it easier to detect
trends, outliers, and correlations, which guides the building of robust statistical
models.

Q4. Write a R-script, load the dataset in ‘R’, display the first six rows and Summarize sales data
to show total sales. (coding).
Ans-
 1. Loading Libraries: The dplyr library is loaded for potential future data manipulation
(if needed).
2. Loading the Dataset: The dataset is loaded using [Link](). Replace
"sales_data.csv" with your dataset file path.
3. Displaying First Six Rows: head(sales_data) shows the first six rows of the dataset to
inspect its structure.
4. Summarizing Total Sales: The sum() function calculates the total sales by summing
up the values in the 'Sales' column. The [Link] = TRUE argument ensures that missing
values (NA) are ignored during the summation.
5. Printing Total Sales: The total sales value is printed using print().
Ensure your dataset has a column named Sales or adjust the column name accordingly in the
script.

UNIT 5 (2 MARKS)
Q1. Define K-means Clustering.
Ans- K-means Clustering is an unsupervised machine learning technique used to group
similar data points into K clusters. It works by repeatedly assigning data points to the closest
cluster and updating the center of each cluster.

Here's a simpler breakdown of the steps:

1. Pick K initial centers (randomly).

2. Assign each data point to the closest center.
3. Update the center of each cluster by finding the average of all the points in that
cluster.
4. Repeat the process until the centers stop changing.

This process helps to organize data into groups that are similar to each other.

Q2. How will use naïve Bayesian Classifier.

Ans-

The Naïve Bayes Classifier is a simple probabilistic machine learning algorithm used for
classification tasks. It is based on Bayes' Theorem, assuming that the features (variables) are
independent of each other.
Here’s how it works in a simple way:

1. Calculate the probability of each class based on the input features.

2. Multiply the individual probabilities of the features given the class.
3. Choose the class with the highest calculated probability.
It’s called "naïve" because it assumes that all features are independent, which may not
always be true, but it still works well for many tasks like spam email detection and
sentiment analysis.

Q3. How association rule can help in Super Market.

Ans-

Association rules in a supermarket help identify patterns in customer purchasing behavior. By

analyzing which products are frequently bought together, the supermarket can:

1. Improve product placement – Place related products near each other (e.g., placing
bread next to butter).
2. Create targeted promotions – Offer discounts or bundle deals on frequently bought-
together items.

This helps increase sales and enhance customer shopping experience by making relevant
suggestions.

Q4. Two real world application of Naïve Bayesian classifier.

Ans-
Two Real-World Applications of Naïve Bayes Classifier:

1. Email Spam Detection: Naïve Bayes is used to classify emails as "spam" or "not
spam" by analyzing the frequency of words in the email. It calculates the probability
of the email being spam based on these word frequencies.
 2. Sentiment Analysis: Naïve Bayes can classify customer reviews or social media
posts as "positive" or "negative" by analyzing the words used in the text and
calculating the likelihood of each sentiment.

Q5. Concept of confidence Metrics in Rule Mining.

Ans-
Confidence is a metric used in association rule mining to evaluate the reliability of a rule. It
measures the likelihood that the rule's conclusion (the consequent) is true given that the rule's premise
(the antecedent) is true.
Significance:
 A high confidence value indicates that when the antecedent occurs, the consequent is likely to
occur as well, making the rule stronger and more useful for decision-making in tasks like
marketing and inventory management.

Q6. Key steps involve in K-Means clustering.

Ans-
The key steps involved in K-means clustering are:
1. Initialization: Select the number of clusters (K) and randomly initialize K centroids
(cluster centers).
2. Assignment: Assign each data point to the nearest centroid based on the Euclidean
distance.
3. Update: Recalculate the centroids by taking the mean of all the points assigned to each
cluster.
4. Repeat: Repeat the assignment and update steps until the centroids no longer change
significantly (convergence).

Q7. Advantages and Disadvantages of Naïve Bayesian Classifier.

Ans-
Advantages of Naïve Bayes Classifier:

1. Simple and Fast: It is easy to implement and very fast, especially with large datasets.
2. Works Well with Small Data: It performs well even with limited training data.
3. Good for Text Classification: It is effective for tasks like spam detection and
sentiment analysis.

Disadvantages of Naïve Bayes Classifier:

1. Assumes Independence: It assumes that features are independent, which may not be
true in real-world data.
2. Poor Performance with Complex Relationships: If features are highly correlated,
the model’s accuracy may drop.
3. Requires Numeric Data: It works best with numeric data and may struggle with
complex categorical variables.
Q8. How No. of Cluster (K) affects the result of k-means algorithm.
Ans-

The number of clusters (K) in the K-means algorithm directly affects how the data is
grouped:

1. Small K (few clusters): If K is too small, the algorithm may group distinct data
points together, losing important details or patterns in the data.
2. Large K (many clusters): If K is too large, the algorithm might overfit the data,
creating too many small clusters that don’t offer meaningful insights, and making the
model more complex than needed.

Choosing the right K is crucial, and methods like the Elbow Method can help find an
optimal value for K by balancing simplicity and detail.

Q9. Evaluate effectiveness in K-means Clustering.

Ans-
The effectiveness of K-means clustering can be evaluated using the following methods:
1. Inertia (Within-Cluster Sum of Squares): This measures the compactness of the clusters. A
lower inertia value indicates that the data points are closer to their centroids, indicating better
clustering performance.
2. Silhouette Score: This evaluates both the cohesion (how close data points in a cluster are)
and separation (how distinct a cluster is from others). A higher silhouette score (closer to 1)
indicates better-defined clusters.

Q10. Discuss the potential ethical consideration.

Ans-
Potential Ethical Considerations in data analytics and machine learning include:
1. Privacy: Ensuring the protection of personal data is crucial. Improper use or storage of
sensitive information can lead to privacy violations and breaches of trust.
2. Bias and Fairness: Data models should be free from bias that could lead to discrimination
against certain groups (e.g., based on gender, race, or socioeconomic status). Ethical practices
involve ensuring fairness and transparency in model predictions to avoid reinforcing
inequalities.

(6 MARKS)

Q1. How company can implement Naïve Bayesian Classifier, steps involved in training models,
How would you classify new customer review.
Ans-
 Steps to Implement Naïve Bayesian Classifier:
1. Data Collection:
 The first step is to collect and prepare the data. In a company, this could be customer
reviews, feedback, or other text data relevant to the classification task (e.g., sentiment
analysis of product reviews, spam detection in emails).
2. Data Preprocessing:
 Text Cleaning: Remove unnecessary characters (e.g., punctuation, stop words), and
normalize the text (e.g., converting to lowercase).
  Feature Extraction: Convert text into features that the model can understand, like
bag-of-words, TF-IDF (Term Frequency-Inverse Document Frequency), or word
embeddings.
3. Splitting the Data:
 Divide the dataset into training and testing sets. Typically, 70-80% of the data is used
for training, and the rest for testing.
4. Model Training:
 Using the training dataset, train the Naïve Bayes classifier. The model will learn the
probabilities of words (or features) occurring in each class (e.g., positive or negative
sentiment, spam or non-spam).
 Example in R: Use the naiveBayes() function from the e1071 package to fit the
model to the training data.
5. Model Evaluation:
 After training the model, evaluate its performance using the test set. Metrics such as
accuracy, precision, recall, and F1-score can be used to assess how well the model
performs.
6. Model Optimization:
 If necessary, fine-tune the model by adjusting parameters or using techniques like
smoothing (e.g., Laplace smoothing) to handle zero-frequency problems.

 Classifying New Customer Reviews:

1. Text Preprocessing:
 Clean the new review by removing irrelevant characters and normalizing the text (e.g.,
converting to lowercase, removing stop words).
2. Feature Extraction:
 Convert the cleaned text into features, typically using methods like bag-of-words or TF-
IDF.
3. Prediction:
 Pass the transformed features of the new review into the trained Naïve Bayes model to
predict the class (e.g., positive/negative sentiment, spam/non-spam).
4. Interpretation:
 The model will output the predicted class label, and based on the model's probabilities,
the company can interpret the result (e.g., if it's a positive or negative review).

Q2. Strength and Weakness of K-means clustering versus Naïve Bayesian Classifier which
method appropriate for Medicare goods, discuss how the result of both model could be
combine for accuracy.
Ans-
Strengths and Weaknesses of K-means Clustering vs. Naïve Bayesian Classifier:
1. K-means Clustering:
 Strength: Simple and Efficient: K-means is easy to implement and computationally
efficient, making it a good choice for large datasets.
 Weakness: Sensitive to K: The number of clusters (K) must be specified beforehand,
and choosing the wrong K can lead to poor clustering results.
 Weakness: Sensitive to Outliers: K-means is affected by outliers, which can skew the
centroids and lead to inaccurate clustering.
2. Naïve Bayesian Classifier:
 Strength: Fast and Scalable: Naïve Bayes is computationally fast and works well
with both categorical and continuous data, making it suitable for large-scale
classification tasks.
 Weakness: Independence Assumption: It assumes that features are independent,
which is often not true in real-world data, reducing the model’s accuracy.
  Weakness: Not Ideal for Complex Relationships: Naïve Bayes struggles with
capturing complex feature relationships, which may reduce its effectiveness in certain
tasks.

Which Method is More Appropriate for Medicare Goods?

3. For Clustering: K-means would be useful for segmenting customers based on purchasing
behavior or grouping similar Medicare goods for better inventory management.
4. For Classification: Naïve Bayes would be suitable for predicting customer behavior, such as
classifying reviews or predicting which customers might purchase specific Medicare products
based on historical data.

Combining Results of Both Models for Accuracy:

5. Ensemble Approach: By combining both models, K-means can group data into clusters, and
then Naïve Bayes can be applied within each cluster to improve classification accuracy.
6. Model Stacking: You can stack predictions from both K-means and Naïve Bayes, treating K-
means cluster labels as an additional feature in the Naïve Bayes classifier to improve
decision-making.
7. Majority Voting: In case of disagreement between the models, a majority voting system can
be used where the most frequent classification between K-means and Naïve Bayes is chosen,
increasing prediction reliability.

Q3. Create a case study to illustrate the use Naïve Bayesian Classifier to classify customer
review Positive, negative and neutral. Brief description of Dataset, features and evaluation
metrics.
Ans-
1. Objective:
 The goal is to classify customer reviews into three categories: Positive, Negative, and
Neutral using the Naïve Bayes classifier.
2. Dataset Description:
 The dataset consists of 10,000 customer reviews collected from an e-commerce
platform, each labeled as Positive, Negative, or Neutral based on sentiment.
3. Features in Dataset:
 The dataset includes Review Text, Product Category, Rating, Review Length, and
Keywords extracted from the reviews.
4. Data Preprocessing:
 Reviews are cleaned by removing stop words, punctuation, and converting text to
lowercase. Features like the Bag of Words or TF-IDF are used for text representation.
5. Splitting the Dataset:
 The data is split into 80% Training and 20% Testing sets to train the model and
evaluate its performance.
6. Model Training:
 The Naïve Bayes model is trained using the review text and other features as input,
with the sentiment labels (Positive, Negative, Neutral) as the output.
7. Model Testing:
 The trained model is tested on the unseen test set, and predictions are made for
sentiment classification.
8. Evaluation Metrics:
 Accuracy: Measures the proportion of correctly classified reviews.
  Precision, Recall, F1-Score: These metrics assess the model’s ability to classify each
sentiment correctly.
9. Confusion Matrix:
 A confusion matrix is used to compare the actual vs. predicted sentiment labels,
showing how well the model distinguishes between different sentiments.
10. Conclusion:
 Naïve Bayes is effective for classifying customer reviews based on sentiment. The
model’s performance can be evaluated using various metrics to ensure reliable
sentiment analysis for business decisions.

Q4. Strength and Weakness of K-means clustering versus Naïve Bayesian Classifier which
method appropriate for Medicare goods, discuss how the result of both model could be
combine for accuracy for high risk customer.
Ans-
 Strengths and Weaknesses of K-means Clustering vs. Naïve Bayesian Classifier:
K-means Clustering:
1. Strength: Unsupervised Learning
 K-means does not require labeled data, which is useful when the exact outcome (e.g.,
customer segments) is unknown. It groups similar items based on features like
demographics or behavior.
2. Strength: Efficient for Large Datasets
 K-means is computationally efficient, making it suitable for large datasets with
numerous customers or products, especially in the case of Medicare goods.
3. Weakness: Sensitive to Initial Centroids
 The performance of K-means can be influenced by the initial placement of centroids.
Poor initial selection can lead to suboptimal clustering results.
4. Weakness: Assumes Spherical Clusters
 K-means assumes that clusters are spherical and evenly sized, which might not be
suitable for more complex, non-uniform data distributions.
 Naïve Bayesian Classifier:
1. Strength: Fast and Simple
 Naïve Bayes is quick to train and easy to implement, making it ideal for classifying
customer data into risk categories based on features like medical history or age.
2. Strength: Handles Both Categorical and Continuous Data
 Naïve Bayes can handle mixed data types (e.g., categorical like gender and
continuous like age), making it flexible for real-world datasets like those in the
healthcare industry.
3. Weakness: Independence Assumption
 Naïve Bayes assumes that features are independent, which may not hold true in
complex healthcare data, potentially reducing model accuracy.
4. Weakness: Limited in Handling Complex Relationships
 The model might struggle with datasets where features interact in complex ways (e.g.,
age, medication, and lifestyle habits affecting health outcomes).

 Which Method is More Appropriate for Medicare Goods?

 K-means Clustering is more appropriate for segmenting customers based on similar
characteristics, such as age, health history, or risk factors. It can be used to group customers
into clusters, such as high-risk vs. low-risk groups for certain Medicare goods or services.
 Naïve Bayes Classifier is better suited for classification tasks, such as predicting whether a
customer is at high risk of certain health conditions or is likely to need a specific product
based on their health profile.
 Combining Both Models for High-Risk Customer Prediction:
1. K-means Clustering for Segmentation:
 Use K-means clustering to segment customers into different groups (e.g., high-risk
and low-risk customers) based on features like age, medical history, and purchasing
behavior.
2. Naïve Bayes for Classification:
 Once the high-risk customers are identified through clustering, apply Naïve Bayes to
classify these high-risk groups based on more specific features like customer
feedback, medication history, or chronic conditions.
3. Combining Results:
 The cluster label from K-means (e.g., high-risk cluster) can be used as an additional
feature in the Naïve Bayes classifier to improve its predictions. This would help the
classifier make more accurate predictions by considering both the segmentation and
individual features.
4. Improved Accuracy:
 By combining the strengths of both models, the overall accuracy of identifying high-
risk customers can be improved. K-means helps with group identification, and Naïve
Bayes ensures that predictions within those groups are accurate, leading to better-
targeted marketing and resource allocation for Medicare products.

UNIT 6 (2 MARKS)

Q1. Difference Between text analytics and sentiment analytics.

Ans-

Q2. Define Linear Regression.

Ans-
Linear Regression is a statistical method used to model the relationship between a dependent
variable (target) and one or more independent variables (predictors) by fitting a linear
equation to observed data. The goal is to find the line (or hyperplane) that best predicts the
dependent variable based on the independent variables.
 Formula: The general equation for linear regression is:
Y=β0+β1X1+β2X2+⋯+βnXn+ϵ
Where:
  Y is the dependent variable,
 X1,X2,…,Xn are independent variables,
 β0 is the intercept,
 β1,β2,…,βn are coefficients of the independent variables,
 ϵ is the error term.

Q3. How can time series analytics to forecast stock prices.

Ans-
Time series analytics can be used to forecast stock prices by analyzing historical data points
(e.g., daily closing prices) over time to identify patterns, trends, and seasonal variations.
Techniques such as ARIMA (Auto Regressive Integrated Moving Average), exponential
smoothing, and machine learning models can be applied to predict future stock prices based on
past behaviors.

1. ARIMA: This model identifies relationships in historical stock price data and forecasts future
prices by considering patterns like trends and seasonality.
2. Exponential Smoothing: This method weighs recent prices more heavily, providing forecasts
that react more quickly to changes in the stock price.

Q4. Difference between Linear and Logistic Regression

Ans-

Q5. Importance of trend and Seasonality in Time Series.

Ans-
1. Trend:
 Trend represents the long-term movement or direction of the data over time. Identifying
the trend helps in understanding the overall growth or decline in the data, such as
increasing sales or decreasing stock prices, which is crucial for making long-term
forecasts.
2. Seasonality:
  Seasonality refers to periodic fluctuations in the data that occur at regular intervals, such
as monthly, quarterly, or annually. Recognizing seasonal patterns allows for better short-
term predictions, like predicting higher sales during the holiday season or weather-
related effects on energy consumption.

Q6. Two key components of time series.

Ans-
1. Trend:
 The long-term movement or direction in the data, indicating whether the values are
generally increasing, decreasing, or remaining constant over time.
2. Seasonality:
 The repeating patterns or fluctuations in the data that occur at regular intervals, such
as daily, monthly, or yearly, often influenced by external factors like weather,
holidays, or economic cycles.

Q7. How would you use time series analysis to forecast future price.
Ans-
To forecast future prices using time series analysis, follow these steps:
1. Collect Historical Data: Gather past price data (e.g., daily stock prices or monthly
sales) to identify trends and patterns.
2. Choose a Model: Use models like ARIMA (Auto-Regressive Integrated Moving
Average) or Exponential Smoothing to analyze the data.
3. Train the Model: Fit the chosen model to the historical data to learn patterns like
seasonality, trends, and cycles.
4. Make Predictions: Use the trained model to predict future prices based on the
identified patterns.
This approach helps in forecasting price changes over time, such as stock prices or product
demand.

Q8. How can time series analytics to forecast stock prices.

Ans-
Time series analytics can be used to forecast stock prices by analyzing historical price data to
identify trends, patterns, and seasonal behaviors. Common methods include:
1. ARIMA (AutoRegressive Integrated Moving Average):
 ARIMA is used to model and predict stock prices by capturing the dependencies in past
data and adjusting for trends and seasonality, making it suitable for stock price forecasting.
2. Exponential Moving Average (EMA)
 EMA is a weighted moving average where more recent data points are given higher
weight, making it more responsive to recent price changes. This is particularly useful for
stock price forecasting because stock prices often exhibit sudden movements that require
quick adjustment in predictions.

Q9. Elaborate Decision Tree.

Ans-
A Decision Tree is a supervised machine learning algorithm used for both classification and
regression tasks. It splits the data into subsets based on feature values to make predictions.
1. Structure: It consists of nodes:
  Root Node: The topmost node that represents the entire dataset.
 Decision Nodes: Internal nodes where the data is split based on a feature condition.
 Leaf Nodes: The final nodes representing the predicted outcome or class label.
2. Process: The tree is built by recursively splitting the dataset at each decision node based on
the feature that best separates the data (using metrics like Gini Index or Information Gain for
classification). The goal is to create a tree that minimizes prediction error.

Q10. Importance of Trend and Seasonality in time series.

Ans- (Q5- same question)

(6 MARKS)

Q1. How would you use linear regression, what insights would you gave as a sports analytics.
Ans-
1. Objective:
 The primary goal of using linear regression in sports analytics is to predict a
continuous outcome, such as player performance, team win probability, or score
predictions based on historical data and player/team statistics.
2. Data Collection:
 Collect historical data on key performance indicators (KPIs), such as player stats
(points, rebounds, assists in basketball), or team stats (goals, possession, shots in
football), alongside the corresponding outcomes (game scores, wins/losses).
3. Feature Selection:
 Select features that are believed to influence the outcome, like a player's average
points per game, shooting percentage, or a team’s defensive efficiency. These
features will be used as the independent variables in the model.
4. Building the Model:
 Use linear regression to model the relationship between the chosen features (e.g.,
player stats) and the target variable (e.g., match outcome or total points scored). The
model will estimate coefficients for each feature, representing their impact on the
predicted outcome.
5. Insight Generation:
 By examining the coefficients in the regression model, insights can be gained on how
different features affect performance. For instance, if the model shows a high
coefficient for "field goal percentage" in predicting winning probability, it suggests
that teams with higher shooting accuracy are more likely to win.
6. Predictions and Strategy:
 Linear regression can be used to predict future outcomes based on current or
projected player stats. These predictions can inform coaching decisions, such as
adjusting lineups, focusing on improving key areas like shooting or defense, or
predicting the outcome of upcoming games.
Example Insights:
 Identifying the key factors influencing a player's performance (e.g., how minutes played, field
goals attempted, and assists correlate with total points).
 Forecasting the total points a team will score based on their offensive stats.
 Understanding how factors like home-field advantage or historical head-to-head records
influence match outcomes.

Q2. Company has both structured and unstructured customer review, which type of
analytics text or sentiment analytics would be more appropriate.
Ans-
 1. Structured Data Overview:
 Structured customer data includes fields like ratings, dates, and demographics. It’s useful
for numerical analysis and trend identification, such as which products have higher
ratings or which customer segments are most satisfied.
2. Unstructured Data Overview:
 Unstructured data consists of free-text customer reviews. Analyzing this data requires
more advanced techniques like text analytics and sentiment analysis to extract
meaningful insights from customer opinions and feedback.
3. Text Analytics for Unstructured Data:
 Text analytics helps process and extract key information from unstructured data, such as
identifying specific topics, keywords, or themes (e.g., quality, service). It categorizes
reviews based on these themes for further analysis.
4. Sentiment Analysis for Emotional Tone:
 Sentiment analysis classifies customer feedback into categories like positive, negative, or
neutral. This helps companies understand overall customer satisfaction and track the
emotional tone of reviews over time.
5. Extracting Trends from Structured Data:
 Text analytics can be applied to structured data by linking customer reviews with product
features, helping identify trends such as which product attributes are most frequently
mentioned in reviews.
6. Understanding Customer Emotions:
 Sentiment analysis enables businesses to detect customer emotions (e.g., frustration,
excitement) in reviews. It helps assess whether customers are satisfied with the product
or service, aiding in customer experience improvement.
7. Combining Text and Sentiment Analytics:
 Combining text and sentiment analysis allows businesses to go beyond just identifying
keywords, helping to understand how those keywords are perceived. For example, the
word "slow" might have a negative sentiment when talking about delivery but neutral
when discussing a feature.

8. Category-based Analysis:
 Using text analytics, reviews can be categorized into groups like "product quality,"
"delivery," or "customer service." Sentiment analysis can then be applied to assess how
customers feel about each specific aspect of the product or service.
9. Targeted Business Decisions:
 The combination of both techniques enables businesses to take targeted actions, such as
improving a product’s quality if many negative sentiments are associated with it, or
enhancing delivery processes if delays are frequently mentioned.
10. Tracking Customer Sentiment Trends:
 Over time, sentiment analysis can track changes in customer sentiment across different
periods, helping businesses monitor how product improvements, promotions, or market
changes affect overall satisfaction and perception.

Q3. Design a detail implementation plan for a bank design tree model, outline entire process,
what method would you recommend.
Ans-
1. Define the Objective:
 The first step is to clearly define what the model should predict, such as loan approval,
credit risk, or customer behavior. For example, predicting whether a loan application will
be approved or rejected based on customer data.
2. Data Collection:
  Gather relevant customer data, such as income, credit score, loan amount, and past
repayment history. This data can come from internal bank records or customer
applications.
3. Data Cleaning and Preprocessing:
 Clean the data by handling missing values, correcting errors, and removing duplicates.
Preprocess categorical data, like marital status, by converting it into numeric form (e.g.,
0 for single, 1 for married).
4. Split Data into Training and Testing:
 Divide the data into two parts: one for training the model (usually 70-80%) and the other
for testing the model’s performance (20-30%).
5. Build the Decision Tree Model:
 Use a CART (Classification and Regression Trees) algorithm to create a decision tree.
The tree splits data into branches based on key features (like credit score) that affect the
prediction outcome.
6. Model Evaluation:
 Evaluate the model's performance using metrics such as accuracy, precision, and recall.
This helps understand how well the model is making predictions (e.g., loan approval or
rejection).
7. Pruning the Tree:
 To avoid overfitting (when the model is too complex), prune the decision tree by
removing unnecessary branches. This makes the model simpler and more general.
8. Interpret the Model:
 Decision trees are easy to interpret. You can directly see the rules, like “If credit score >
700, and income > $50,000, approve the loan,” making the model transparent and
understandable for bank staff.
9. Deploy the Model:
 Once the model is trained and tested, it can be integrated into the bank’s system for real-
time decision-making. For instance, loan officers can use the model to automatically
approve or reject loan applications.

10. Monitor and Update:

 Continuously monitor the model’s performance and retrain it with new data to ensure it
remains accurate. Periodically update the model to account for changes in customer
behavior or market conditions.

Q4. Define sentiment analytics, describe its primary components, what types of data R typically
analyze in sentiment analysis.
Ans-
1. Definition:
 Sentiment analytics refers to analyzing text data to determine the sentiment or emotional tone
expressed. It categorizes sentiments into positive, negative, or neutral, helping businesses
understand public perception.
2. Text Preprocessing:
 The first step in sentiment analysis is cleaning the text. This involves removing irrelevant
information like stop words, punctuation, and special characters, and normalizing the text
through stemming or lemmatization.
3. Feature Extraction:
 In this step, key features such as important words, phrases, or word frequency are extracted
from the text. Methods like TF-IDF (Term Frequency-Inverse Document Frequency) are
used to weigh words based on their significance.
4. Sentiment Classification:
  Sentiment classification is done by applying machine learning models such as Naïve Bayes or
Support Vector Machines (SVM) to categorize the sentiment of the text as positive,
negative, or neutral.
5. Polarity Scoring:
 Sentiment analysis often includes polarity scoring, where the sentiment's intensity is
quantified. Positive polarity indicates a positive sentiment, negative polarity represents a
negative sentiment, and neutral polarity means no strong emotion.
6. Types of Data Analyzed in R:
 R can analyze various types of data for sentiment analysis, including customer reviews,
social media posts, survey responses, and news articles, extracting sentiment from text to
gain insights into customer opinions or market trends.
7. R Libraries for Sentiment Analysis:
 R offers several packages for sentiment analysis, including tm for text mining, tidytext for
text processing, and syuzhet for sentiment scoring, providing tools for efficient text analysis.
8. Applications in Business:
 Sentiment analytics helps businesses understand customer feedback, assess brand reputation,
and track public opinion on various topics. It’s widely used in marketing, customer service,
and product development.