1.

1 Data sources are identified that are relevant to the business problem

This means figuring out where the right data will come from so the business problem can
actually be analyzed and solved.

What it involves

●​ Understanding the business objective (e.g., reduce customer churn, increase sales).
●​ Deciding what data is needed to measure, explain, or predict the problem.
●​ Identifying reliable and appropriate data sources that contain this data.

Common types of data sources

●​ Internal sources
○​ Transaction databases (sales, orders, payments)
○​ Customer records (CRM systems)
○​ Operational data (logs, inventory, call center data)
●​ External sources
○​ Government or public datasets
○​ Market research reports
○​ Social media data
○​ Third-party data providers
●​ Primary data
○​ Surveys
○​ Interviews
○​ Experiments
●​ Secondary data
○​ Previously collected datasets
○​ Reports and historical records

Example

Business problem: A telecom company wants to reduce customer churn​

Relevant data sources:

●​ Customer usage data (call duration, internet usage)

●​ Billing and payment history
●​ Customer complaints and service tickets
●​ Demographic data
Why this is important

●​ Using irrelevant data leads to wrong conclusions.

●​ Missing key data can make the analysis incomplete or biased.
●​ Good data source identification saves time, cost, and effort later in the project.

Sure 👍 ​
Here are MCQ questions only (no answers) for 1.1 Data sources are identified that are
relevant to the business problem — exam-focused.

📘 MCQs: Identifying Relevant Data Sources

1.​ What is the main objective of identifying data sources in a data science project?​
A. To clean the data​
B. To select data relevant to the business problem​
C. To build machine learning models​
D. To visualize results
2.​ Which phase of a data science project focuses on identifying relevant data sources?​
A. Data Preparation​
B. Modeling​
C. Business Understanding​
D. Evaluation
3.​ Which of the following is an example of an internal data source?​
A. Government census data​
B. Social media posts​
C. Company sales database​
D. Market research report
4.​ External data sources are best described as data that:​
A. Is stored in company servers only​
B. Is collected directly from employees​
C. Comes from outside the organization​
D. Is always unstructured
5.​ Which data source would be MOST relevant for analyzing customer churn in a telecom
company?​
A. Weather reports​
B. Customer usage and billing data​
C. Stock market prices​
D. Employee attendance records
6.​ Primary data refers to data that is:​
A. Purchased from third parties​
B. Collected directly for a specific purpose​
C. Taken from historical records​
D. Always publicly available
 7.​ Which of the following is a secondary data source?​
A. Customer satisfaction survey​
B. Interviews conducted by analysts​
C. Previously collected company reports​
D. Field experiments
8.​ Why is identifying irrelevant data sources harmful to a project?​
A. It increases storage capacity​
B. It leads to faster modeling​
C. It can result in incorrect conclusions​
D. It improves data quality
9.​ Which factor should be considered while selecting a data source?​
A. Data relevance​
B. Data accuracy​
C. Data availability​
D. All of the above
10.​Social media data is usually classified as:​
A. Internal data​
B. Primary data​
C. External data​
D. Transactional data

✅
Here are the answers to the MCQs on 1.1 Data sources are identified that are relevant to
the business problem

📘 MCQ Answers
1.​ B – To select data relevant to the business problem
2.​ C – Business Understanding
3.​ C – Company sales database
4.​ C – Comes from outside the organization
5.​ B – Customer usage and billing data
6.​ B – Collected directly for a specific purpose
7.​ C – Previously collected company reports
8.​ C – It can result in incorrect conclusions
9.​ D – All of the above
10.​C – External data
Perfect 👍 ​
I’ll do this in order, exactly as you asked:

1.2 Data is loaded from multiple data sources –

Explanation
This means collecting and importing data from more than one source into a single system
or environment for analysis.

In real-world business problems, data rarely comes from one place. For example, customer data
may be stored in a database, sales data in Excel files, and feedback data from online APIs.
All these datasets must be loaded together so they can be analyzed collectively.

Key points:

●​ Data may come from databases, files, APIs, cloud storage, or sensors
●​ Data loading is done using ETL tools or programming languages like Python or SQL
●​ Combining multiple sources gives a complete and accurate view of the business
problem
●​ Challenges include different formats, missing values, and inconsistent data

Example:

Business Problem: Improve customer satisfaction​

Data Sources Loaded:

●​ Customer database (CRM)

●​ Survey results (CSV files)
●​ Social media feedback (API)

MCQs: 1.2 Data is loaded from multiple data sources

1.​ What does loading data from multiple data sources involve?​
A. Exporting data to one file​
B. Importing data from different sources into a system​
C. Cleaning data only​
D. Visualizing data
2.​ Why is data often loaded from multiple sources?​
A. To increase data duplication​
 B. To get a complete picture of the business problem​
C. To reduce analysis time​
D. To avoid data integration
3.​ Which of the following represents multiple data sources?​
A. Single CSV file​
B. Only one database table​
C. Database, API, and Excel file​
D. One text document
4.​ Which tool is commonly used to load data from different sources?​
A. Compiler​
B. ETL tool​
C. Text editor​
D. Web browser
5.​ What is a common challenge when loading data from multiple sources?​
A. Perfect data consistency​
B. Data inconsistency​
C. Automatic modeling​
D. Faster performance
6.​ After loading data from multiple sources, which step is usually performed next?​
A. Business understanding​
B. Data cleaning and preprocessing​
C. Model deployment​
D. Final reporting
7.​ Data from CRM systems and transaction databases together indicate:​
A. Single-source data​
B. External data​
C. Multi-source data​
D. Synthetic data
8.​ Which file format is commonly used when loading data from different sources?​
A. CSV​
B. MP3​
C. JPG​
D. EXE
9.​ Why is data integration necessary after loading multiple datasets?​
A. To delete all data​
B. To combine datasets meaningfully​
C. To reduce storage​
D. To encrypt data
10.​Loading data from APIs, flat files, and databases is an example of:​
A. Homogeneous data loading​
B. Single-source data loading​
C. Multi-source data loading​
D. Manual data entry
Here are the answers for 1.2 Data is loaded from multiple data sources ✅

📘 MCQ Answer Key

1.​ B – Importing data from different sources into a system
2.​ B – To get a complete picture of the business problem
3.​ C – Database, API, and Excel file
4.​ B – ETL tool
5.​ B – Data inconsistency
6.​ B – Data cleaning and preprocessing
7.​ C – Multi-source data
8.​ A – CSV
9.​ B – To combine datasets meaningfully
10.​C – Multi-source data loading

1.3 Complex data is loaded using appropriate data

acquisition techniques – Explanation
Complex data refers to data that is large, unstructured, semi-structured, or generated at
high speed. Examples include images, videos, audio, text, sensor data, social media data, and
streaming data.

Because such data cannot be handled using simple file-loading methods, special data
acquisition techniques are required to collect and load it efficiently and accurately.

Key points:

●​ Complex data may be unstructured (text, images, videos) or semi-structured (JSON,

XML)
●​ Requires advanced acquisition techniques like APIs, web scraping, streaming tools,
and big data frameworks
●​ Tools such as APIs, web crawlers, Hadoop, Spark, and IoT pipelines are commonly
used
●​ Ensures data is collected in a scalable, reliable, and real-time manner when required
Examples:

●​ Social media posts collected using APIs

●​ Sensor data collected using IoT pipelines
●​ Log data processed using distributed systems

MCQs: 1.3 Complex data is loaded using appropriate data

acquisition techniques
1.​ Which of the following best describes complex data?​
A. Small and well-structured data​
B. Data stored only in Excel files​
C. Unstructured or high-volume data​
D. Numeric data only
2.​ Which data type is considered unstructured?​
A. Relational tables​
B. CSV files​
C. Text and images​
D. Spreadsheets
3.​ JSON and XML files are examples of:​
A. Structured data​
B. Unstructured data​
C. Semi-structured data​
D. Numeric data
4.​ Which technique is MOST suitable for acquiring social media data?​
A. Manual data entry​
B. APIs​
C. Keyboard input​
D. Text editors
5.​ Which tool is commonly used for processing large-scale complex data?​
A. MS Excel​
B. Notepad​
C. Apache Spark​
D. Calculator
6.​ Streaming data is best acquired using:​
A. Static file download​
B. Batch processing only​
C. Real-time data pipelines​
D. Printed reports
 7.​ Web scraping is mainly used to collect:​
A. Sensor data​
B. Data from websites​
C. Audio signals​
D. Database logs
8.​ Which challenge is common when dealing with complex data?​
A. Simple structure​
B. High consistency​
C. Scalability issues​
D. Low data volume
9.​ Which data acquisition method is suitable for IoT sensor data?​
A. Surveys​
B. APIs and streaming pipelines​
C. Manual forms​
D. Spreadsheets
10.​The main purpose of using appropriate data acquisition techniques is to:​
A. Reduce data accuracy​
B. Handle complex data efficiently​
C. Avoid data storage​
D. Eliminate preprocessing

✅
Here are the answers for 1.3 Complex data is loaded using appropriate data acquisition
techniques

📘 MCQ Answer Key

1.​ C – Unstructured or high-volume data
2.​ C – Text and images
3.​ C – Semi-structured data
4.​ B – APIs
5.​ C – Apache Spark
6.​ C – Real-time data pipelines
7.​ B – Data from websites
8.​ C – Scalability issues
9.​ B – APIs and streaming pipelines
10.​B – Handle complex data efficiently
Perfect 👍 ​
I’ll follow the same clean pattern as before: explanation first, then MCQs for both 2.1 and
2.2.

2.1 Cases which require data transformations are

identified – Explanation
Data transformation is required when raw data is not suitable for analysis or modeling in its
original form. Identifying such cases helps improve model accuracy, consistency, and
performance.

Common cases where data transformation is required:

●​ Different scales of features (e.g., age vs income) → normalization/standardization

●​ Categorical variables → encoding (label encoding, one-hot encoding)
●​ Skewed data → log or power transformation
●​ Missing values → imputation
●​ Non-linear relationships → mathematical transformations

Why this is important:

●​ Ensures data meets model assumptions

●​ Improves learning efficiency of algorithms
●​ Makes features comparable and meaningful

Example:

If income ranges from 10,000–1,000,000 and age ranges from 18–80, scaling is required to
avoid bias in models.

2.2 Impacts of imbalanced data are explained –

Explanation
Imbalanced data occurs when one class significantly outnumbers another, commonly seen
in classification problems like fraud detection or disease diagnosis.
Impacts of imbalanced data:

●​ Model becomes biased toward the majority class

●​ High overall accuracy but poor minority class prediction
●​ Important cases (fraud, disease) may be missed
●​ Misleading evaluation metrics

Example:

If 95% transactions are normal and 5% are fraud, a model predicting “normal” always will still
get 95% accuracy but is useless.

MCQs: 2.1 Data Transformation

1.​ Data transformation is required when:​
A. Data is already clean​
B. Features are on different scales​
C. Data is perfectly balanced​
D. Data is visualized
2.​ Which transformation is used to handle skewed data?​
A. Encoding​
B. Normalization​
C. Log transformation​
D. Sampling
3.​ Converting categorical data into numerical form is called:​
A. Scaling​
B. Encoding​
C. Normalization​
D. Aggregation
4.​ Which situation MOST requires data transformation?​
A. All features have same range​
B. No missing values​
C. Different units and scales in data​
D. Small dataset size
5.​ Standardization is mainly used to:​
A. Remove duplicates​
B. Scale data to mean 0 and variance 1​
C. Balance classes​
D. Reduce dimensions
MCQs: 2.2 Imbalanced Data
6.​ Imbalanced data refers to:​
A. Equal number of classes​
B. Missing values in data​
C. Unequal class distribution​
D. Noisy data
7.​ A major impact of imbalanced data is:​
A. Faster training​
B. Model bias toward majority class​
C. Improved minority prediction​
D. Reduced accuracy always
8.​ Which metric is misleading in imbalanced datasets?​
A. Precision​
B. Recall​
C. Accuracy​
D. F1-score
9.​ Imbalanced data is COMMON in which problem?​
A. Weather forecasting​
B. Fraud detection​
C. Linear regression​
D. Image resizing
10.​A model with 99% accuracy but zero fraud detection indicates:​
A. Good performance​
B. Balanced data​
C. Class imbalance issue​
D. Overfitting solved

✅
Here are the answers for 2.1 Cases which require data transformations and 2.2 Impacts of
imbalanced data

📘 MCQ Answer Key

2.1 Data Transformation

1.​ B – Features are on different scales

2.​ C – Log transformation
3.​ B – Encoding
4.​ C – Different units and scales in data
5.​ B – Scale data to mean 0 and variance 1
2.2 Imbalanced Data

6.​ C – Unequal class distribution

7.​ B – Model bias toward majority class
8.​ C – Accuracy
9.​ B – Fraud detection
10.​C – Class imbalance issue

Alright 👍 ​
Here’s 2.3 Feature selection techniques are applied — first the explanation, then MCQs,
same exam-ready style.

2.3 Feature selection techniques are applied –

Explanation
Feature selection is the process of choosing the most relevant features (variables) from a
dataset for building a machine learning or statistical model.

Not all features contribute equally. Some may be irrelevant, redundant, or noisy, which can
reduce model performance. Feature selection helps improve accuracy, efficiency, and
interpretability.

Why feature selection is important:

●​ Reduces overfitting
●​ Improves model performance
●​ Decreases training time
●​ Enhances model interpretability

Common feature selection techniques:

●​ Filter methods – correlation, chi-square test, ANOVA

●​ Wrapper methods – forward selection, backward elimination
●​ Embedded methods – Lasso (L1), Ridge (L2), decision trees

Example:
In predicting house prices, features like location and size may be useful, while house ID is
irrelevant and should be removed.

MCQs: 2.3 Feature Selection Techniques

1.​ What is the main purpose of feature selection?​
A. Increase dataset size​
B. Remove irrelevant or redundant features​
C. Convert categorical data​
D. Balance the dataset
2.​ Which of the following is a filter-based feature selection method?​
A. Forward selection​
B. Lasso regression​
C. Correlation analysis​
D. Decision trees
3.​ Wrapper methods select features by:​
A. Using statistical tests only​
B. Evaluating feature subsets using a model​
C. Random selection​
D. Scaling features
4.​ Which embedded method performs feature selection during model training?​
A. Chi-square test​
B. ANOVA​
C. Lasso regression​
D. Correlation matrix
5.​ Feature selection helps reduce:​
A. Bias​
B. Overfitting​
C. Data imbalance​
D. Missing values
6.​ Which technique removes one feature at a time starting from all features?​
A. Forward selection​
B. Backward elimination​
C. Random sampling​
D. Normalization
7.​ Highly correlated features should be:​
A. Always kept​
B. Randomly removed​
C. Reduced to avoid redundancy​
D. Ignored
8.​ Feature selection mainly improves which of the following?​
A. Model interpretability​
 B. Data volume​
C. Data imbalance​
D. Noise generation
9.​ Which of the following is NOT a feature selection technique?​
A. PCA​
B. Forward selection​
C. Backward elimination​
D. Lasso regression
10.​Removing irrelevant features usually results in:​
A. Slower training​
B. Poor performance​
C. Better model accuracy​
D. Data loss always

Here are the answers for 2.3 Feature selection techniques are applied ✅

📘 MCQ Answer Key

1.​ B – Remove irrelevant or redundant features
2.​ C – Correlation analysis
3.​ B – Evaluating feature subsets using a model
4.​ C – Lasso regression
5.​ B – Overfitting
6.​ B – Backward elimination
7.​ C – Reduced to avoid redundancy
8.​ A – Model interpretability
9.​ A – PCA
10.​C – Better model accuracy

👍
Great, let’s continue ​
Below is 2.4 Data transformation is performed — first the explanation, then MCQs, in the
same exam-ready format.

2.4 Data transformation is performed – Explanation

Data transformation is the process of converting data from its original form into a suitable
format for analysis or modeling. After identifying the need for transformation (in earlier steps),
the actual transformations are applied.

Common data transformation operations:

●​ Scaling – normalization, standardization

●​ Encoding – label encoding, one-hot encoding
●​ Handling skewness – log, square root transformation
●​ Aggregation – summarizing data
●​ Discretization – converting continuous data into categories

Why data transformation is performed:

●​ To meet algorithm assumptions

●​ To make features comparable
●​ To improve model performance and convergence
●​ To reduce bias caused by scale or distribution

Example:

Applying standardization to numerical features before using k-means or logistic regression.

MCQs: 2.4 Data transformation is performed

1.​ Data transformation is performed to:​
A. Increase data size​
B. Convert data into a suitable form for modeling​
C. Delete important features​
D. Balance class distribution
2.​ Which technique is used to scale features between 0 and 1?​
A. Standardization​
B. Normalization​
C. Encoding​
D. Sampling
3.​ Label encoding is mainly applied to:​
A. Numerical variables​
B. Continuous variables​
C. Categorical variables​
D. Target variables only
4.​ Which transformation is used to reduce skewness in data?​
A. One-hot encoding​
 B. Log transformation​
C. Feature selection​
D. Sampling
5.​ Standardization transforms data to have:​
A. Range 0 to 1​
B. Mean 0 and standard deviation 1​
C. Only positive values​
D. Binary values
6.​ Which algorithm is sensitive to feature scaling?​
A. Decision Tree​
B. K-Means clustering​
C. Random Forest​
D. Naive Bayes
7.​ Discretization is the process of:​
A. Removing outliers​
B. Encoding categories​
C. Converting continuous variables into intervals​
D. Scaling values
8.​ Performing data transformation mainly improves:​
A. Data collection​
B. Model performance​
C. Data imbalance​
D. Storage cost
9.​ One-hot encoding converts a categorical variable into:​
A. Single numeric column​
B. Multiple binary columns​
C. Text labels​
D. Continuous values
10.​Data transformation is typically performed in which phase?​
A. Business Understanding​
B. Data Preparation​
C. Model Deployment​
D. Evaluation

Here are the answers for 2.4 Data transformation is performed ✅

📘 MCQ Answer Key

1.​ B – Convert data into a suitable form for modeling
2.​ B – Normalization
3.​ C – Categorical variables
 4.​ B – Log transformation
5.​ B – Mean 0 and standard deviation 1
6.​ B – K-Means clustering
7.​ C – Converting continuous variables into intervals
8.​ B – Model performance
9.​ B – Multiple binary columns
10.​B – Data Preparation

Perfect 👍 ​
I’ll keep the same exam-friendly flow: clear explanation first, then MCQs for both 2.5 and
2.6.

2.5 Slicing, indexing, sub-setting, merging and joining

datasets are performed – Explanation
These operations are used to access, manipulate, and combine data during data preparation
and analysis.

Key concepts:

●​ Indexing: Selecting data using row/column labels or positions

●​ Slicing: Selecting a range of rows or columns
●​ Sub-setting: Extracting a specific part of the dataset based on conditions
●​ Merging: Combining datasets based on common columns (similar to SQL JOIN)
●​ Joining: Combining datasets using indexes or keys

Why this is important:

●​ Enables focused analysis

●​ Helps combine data from multiple tables
●​ Essential for data cleaning and feature engineering

Example:

●​ Selecting sales data for a specific year (slicing)

●​ Combining customer and order tables using customer_id (merge/join)
2.6 Facts, Dimensions and Schemas are identified and
explained – Explanation
These concepts come from data warehousing and analytics and are used to organize data for
reporting and analysis.

Key terms:

●​ Facts: Numeric, measurable data (e.g., sales amount, quantity sold)

●​ Dimensions: Descriptive attributes that give context (e.g., time, product, customer)
●​ Schema: The structure that defines how facts and dimensions are organized

Common schemas:

●​ Star schema: One fact table connected to multiple dimension tables

●​ Snowflake schema: Dimensions are further normalized into sub-dimension tables

Why this matters:

●​ Improves query performance

●​ Makes data easy to understand
●​ Supports business intelligence and reporting

MCQs: 2.5 Dataset Operations

1.​ Indexing is mainly used to:​
A. Combine datasets​
B. Select data using labels or positions​
C. Store data​
D. Visualize data
2.​ Slicing allows users to:​
A. Remove columns​
B. Select a range of rows or columns​
C. Rename features​
D. Balance data
3.​ Sub-setting data usually involves:​
A. Aggregation only​
B. Conditional selection​
C. Model training​
D. Feature scaling
 4.​ Merging datasets is similar to which SQL operation?​
A. SELECT​
B. UPDATE​
C. JOIN​
D. DELETE
5.​ Joining datasets typically requires:​
A. Same number of rows​
B. Common key or index​
C. Same file format​
D. Balanced classes

MCQs: 2.6 Facts, Dimensions and Schemas

6.​ Facts in a data warehouse are usually:​
A. Textual descriptions​
B. Categorical values​
C. Numeric and measurable data​
D. Metadata
7.​ Which of the following is a dimension?​
A. Sales amount​
B. Profit​
C. Time​
D. Revenue
8.​ A fact table primarily contains:​
A. Only dimension attributes​
B. Measurable business data​
C. Unstructured data​
D. Image files
9.​ In a star schema:​
A. Dimensions are normalized​
B. Fact table is at the center​
C. No dimension tables exist​
D. Only one dimension is allowed
10.​Snowflake schema differs from star schema because it:​
A. Has no fact table​
B. Uses denormalized dimensions​
C. Has normalized dimension tables​
D. Is used only for OLTP
 ✅
Here are the answers for 2.5 Slicing, indexing, sub-setting, merging and joining datasets
and 2.6 Facts, Dimensions and Schemas

📘 MCQ Answer Key

2.5 Dataset Operations

1.​ B – Select data using labels or positions

2.​ B – Select a range of rows or columns
3.​ B – Conditional selection
4.​ C – JOIN
5.​ B – Common key or index

2.6 Facts, Dimensions and Schemas

6.​ C – Numeric and measurable data

7.​ C – Time
8.​ B – Measurable business data
9.​ B – Fact table is at the center
10.​C – Has normalized dimension tables

Alright 👍​
Here’s 2.7 Techniques are applied to handle missing values — first a clear explanation,
then exam-ready MCQs, same pattern as before.

2.7 Techniques are applied to handle missing values –

Explanation
Missing values occur when data is not recorded or unavailable for certain observations.
Handling missing values is essential because many statistical and machine learning algorithms
cannot work properly with missing data.

Common reasons for missing values:

●​ Data entry errors

 ●​ Sensor or system failures
●​ Non-response in surveys
●​ Data corruption

Techniques to handle missing values:

●​ Deletion
○​ Row-wise deletion (drop records with missing values)
○​ Column-wise deletion (drop features with too many missing values)
●​ Imputation
○​ Mean, median, or mode imputation
○​ Forward fill / backward fill
○​ Regression or KNN imputation
●​ Flagging
○​ Creating a new indicator variable to mark missing values

Why this is important:

●​ Prevents model errors

●​ Reduces bias
●​ Improves data quality and model performance

Example:

Replacing missing age values with the median age of the dataset.

MCQs: 2.7 Handling Missing Values

1.​ Missing values occur when:​
A. Data is duplicated​
B. Data is not recorded for some observations​
C. Data is perfectly clean​
D. Data is normalized
2.​ Which technique removes records containing missing values?​
A. Imputation​
B. Encoding​
C. Deletion​
D. Scaling
3.​ Mean and median replacement are examples of:​
A. Feature selection​
B. Data deletion​
 C. Imputation​
D. Sampling
4.​ Which imputation method is best for skewed numerical data?​
A. Mean​
B. Median​
C. Mode​
D. Random
5.​ Forward fill is mainly used in:​
A. Image data​
B. Time-series data​
C. Text data​
D. Categorical encoding
6.​ Dropping a column is appropriate when:​
A. Few missing values exist​
B. Column is highly important​
C. Column has a very high percentage of missing values​
D. Data is balanced
7.​ Creating a missing-value indicator helps to:​
A. Remove all missing data​
B. Increase noise​
C. Preserve information about missingness​
D. Normalize data
8.​ Which problem may occur if missing values are not handled?​
A. Improved accuracy​
B. Model failure or bias​
C. Faster execution​
D. Balanced data
9.​ Mode imputation is most suitable for:​
A. Continuous data​
B. Numerical data​
C. Categorical data​
D. Image data
10.​Handling missing values is mainly performed in which phase?​
A. Business Understanding​
B. Data Preparation​
C. Deployment​
D. Monitoring

Here are the answers for 2.7 Techniques are applied to handle missing values ✅
📘 MCQ Answer Key
1.​ B – Data is not recorded for some observations
2.​ C – Deletion
3.​ C – Imputation
4.​ B – Median
5.​ B – Time-series data
6.​ C – Column has a very high percentage of missing values
7.​ C – Preserve information about missingness
8.​ B – Model failure or bias
9.​ C – Categorical data
10.​B – Data Preparation

Nice catch 👍 we’ll do 2.8 now — explanation first, then MCQs, exactly like the others.

2.8 Outliers are identified, visualized and dealt with –

Explanation
Outliers are data points that significantly differ from other observations in a dataset. They
may occur due to data entry errors, measurement errors, or genuine extreme values.

Identifying and handling outliers is important because they can distort statistical analysis and
negatively affect model performance, especially in algorithms sensitive to extreme values.

How outliers are identified:

●​ Statistical methods
○​ Z-score
○​ IQR (Interquartile Range)
●​ Visualization techniques
○​ Box plots
○​ Scatter plots
○​ Histograms

How outliers are dealt with:

●​ Removing outliers (if they are errors)

●​ Capping or winsorizing values
●​ Transforming data (log transformation)
●​ Keeping outliers (if they are meaningful)
Why this is important:

●​ Improves model accuracy

●​ Reduces skewness
●​ Prevents biased results

Example:

In a salary dataset, a value of 10 million among mostly 30k–100k may be an outlier.

MCQs: 2.8 Outliers

1.​ An outlier is a data point that:​
A. Is missing​
B. Is very different from other observations​
C. Is duplicated​
D. Is normalized
2.​ Which method is commonly used to detect outliers statistically?​
A. Encoding​
B. Z-score​
C. Sampling​
D. Normalization
3.​ The IQR method identifies outliers that lie:​
A. Between Q1 and Q3​
B. Inside the box​
C. Below Q1 − 1.5×IQR or above Q3 + 1.5×IQR​
D. At the mean
4.​ Which visualization is MOST suitable for detecting outliers?​
A. Line chart​
B. Box plot​
C. Pie chart​
D. Bar chart
5.​ Scatter plots are useful for detecting outliers in:​
A. One variable​
B. Two variables​
C. Text data​
D. Categorical data
6.​ Which technique is used to reduce the effect of extreme outliers without removing them?​
A. Deletion​
B. Encoding​
C. Winsorization​
D. Indexing
 7.​ When should an outlier be kept in the dataset?​
A. Always removed​
B. When it is a genuine and meaningful observation​
C. When it increases noise​
D. When it is missing
8.​ Outliers mainly affect which models the most?​
A. Decision trees​
B. Linear regression​
C. Random forest​
D. Naive Bayes
9.​ Log transformation helps in handling outliers by:​
A. Removing data​
B. Increasing variance​
C. Reducing skewness​
D. Duplicating values
10.​Identifying and handling outliers is part of which phase?​
A. Business Understanding​
B. Data Preparation​
C. Deployment​
D. Monitoring

Here are the answers for 2.8 Outliers are identified, visualized and dealt with ✅

📘 MCQ Answer Key

1.​ B – Is very different from other observations
2.​ B – Z-score
3.​ C – Below Q1 − 1.5×IQR or above Q3 + 1.5×IQR
4.​ B – Box plot
5.​ B – Two variables
6.​ C – Winsorization
7.​ B – When it is a genuine and meaningful observation
8.​ B – Linear regression
9.​ C – Reducing skewness
10.​B – Data Preparation
Alright 👍 ​
Here’s 2.9 Fully usable dataset is constructed by cleaning and transforming data — clear
explanation first, then exam-ready MCQs, same style as before.

2.9 Fully usable dataset is constructed by cleaning and

transforming data – Explanation
A fully usable dataset is one that is clean, consistent, complete, and properly formatted so it
can be directly used for analysis or model building.

This step combines all earlier data preparation activities such as handling missing values,
removing duplicates, correcting errors, transforming variables, and selecting relevant
features.

Key activities involved:

●​ Cleaning
○​ Handling missing values
○​ Removing duplicates
○​ Correcting inconsistent or invalid data
●​ Transforming
○​ Scaling and normalization
○​ Encoding categorical variables
○​ Feature selection and creation
●​ Validation
○​ Checking data types
○​ Ensuring data quality and consistency

Why this is important:

●​ Ensures accurate and reliable results

●​ Prevents model errors
●​ Improves model performance
●​ Makes data ready for analysis and modeling

Example:

A customer dataset after removing duplicates, filling missing ages, encoding gender, and
scaling income becomes a fully usable dataset.
MCQs: 2.9 Fully usable dataset construction
1.​ A fully usable dataset is one that is:​
A. Large in size​
B. Clean and ready for modeling​
C. Unprocessed​
D. Raw and unstructured
2.​ Which activity is NOT part of data cleaning?​
A. Removing duplicates​
B. Handling missing values​
C. Encoding categorical variables​
D. Correcting inconsistent data
3.​ Data transformation mainly involves:​
A. Data collection​
B. Scaling and encoding​
C. Data deletion only​
D. Visualization
4.​ Removing duplicate records helps to:​
A. Increase dataset size​
B. Improve data quality​
C. Increase noise​
D. Cause bias
5.​ Encoding categorical variables is done to:​
A. Remove categories​
B. Convert categories into numerical form​
C. Balance classes​
D. Reduce dataset size
6.​ A dataset is considered fully usable when it:​
A. Contains raw data​
B. Has missing and inconsistent values​
C. Is cleaned and transformed​
D. Is unstructured
7.​ Feature selection is performed to:​
A. Add noise​
B. Reduce irrelevant features​
C. Increase redundancy​
D. Duplicate variables
8.​ Scaling features is important to:​
A. Increase feature count​
B. Make features comparable​
C. Remove outliers​
D. Balance data
 9.​ Data validation ensures:​
A. Data is collected​
B. Data is stored​
C. Data quality and consistency​
D. Data is visualized
10.​Constructing a fully usable dataset is mainly part of:​
A. Business Understanding​
B. Data Preparation​
C. Model Deployment​
D. Evaluation

✅
Here are the answers for 2.9 Fully usable dataset is constructed by cleaning and
transforming data

📘 MCQ Answer Key

1.​ B – Clean and ready for modeling
2.​ C – Encoding categorical variables
3.​ B – Scaling and encoding
4.​ B – Improve data quality
5.​ B – Convert categories into numerical form
6.​ C – Is cleaned and transformed
7.​ B – Reduce irrelevant features
8.​ B – Make features comparable
9.​ C – Data quality and consistency
10.​B – Data Preparation

Great👍 ​
Here’s 3.1 Steps of Exploratory Data Analysis (EDA) are described — clear explanation
first, then exam-ready MCQs, same style you’ve been using.

3.1 Steps of Exploratory Data Analysis (EDA) –

Explanation
Exploratory Data Analysis (EDA) is the process of understanding, summarizing, and
exploring data before applying statistical models or machine learning algorithms.
EDA helps identify patterns, trends, relationships, missing values, and outliers in the
dataset.

🔹 Main steps of EDA:

1.​ Understand the data
○​ Check dataset size, structure, and data types
○​ Identify numerical and categorical variables
2.​ Data quality check
○​ Identify missing values
○​ Detect duplicate records
○​ Check inconsistent or invalid data
3.​ Univariate analysis
○​ Analyze one variable at a time
○​ Use summary statistics (mean, median, mode)
○​ Visualizations: histograms, box plots
4.​ Bivariate / Multivariate analysis
○​ Analyze relationships between variables
○​ Correlation analysis
○​ Visualizations: scatter plots, heatmaps
5.​ Outlier detection
○​ Identify extreme values
○​ Use box plots, IQR, Z-score
6.​ Insights and conclusions
○​ Summarize findings
○​ Identify important features
○​ Guide further data preparation or modeling

🔹 Why EDA is important:

●​ Helps make data-driven decisions
●​ Prevents wrong assumptions
●​ Improves model selection and performance

MCQs: 3.1 Steps of EDA

1.​ The main purpose of Exploratory Data Analysis is to:​
A. Build models​
B. Deploy models​
C. Understand and explore data​
D. Store data
2.​ Which step of EDA involves checking data types and dataset shape?​
A. Outlier detection​
B. Data understanding​
C. Visualization​
D. Modeling
3.​ Identifying missing values is part of:​
A. Data collection​
B. Data quality check​
C. Model evaluation​
D. Deployment
4.​ Analyzing a single variable at a time is called:​
A. Bivariate analysis​
B. Multivariate analysis​
C. Univariate analysis​
D. Predictive analysis
5.​ Which visualization is commonly used in univariate analysis?​
A. Scatter plot​
B. Heatmap​
C. Histogram​
D. Pair plot
6.​ Studying the relationship between two variables is known as:​
A. Univariate analysis​
B. Data cleaning​
C. Bivariate analysis​
D. Feature scaling
7.​ Correlation analysis is mainly used in:​
A. Data storage​
B. Bivariate analysis​
C. Data collection​
D. Deployment
8.​ Which EDA step helps identify extreme values?​
A. Data formatting​
B. Outlier detection​
C. Encoding​
D. Sampling
9.​ Box plots are mainly used in EDA to:​
A. Encode data​
B. Detect outliers​
C. Normalize data​
D. Balance classes
 10.​The final step of EDA focuses on:​
A. Data deletion​
B. Model training​
C. Drawing insights and conclusions​
D. Feature encoding

Here are the answers for 3.1 Steps of Exploratory Data Analysis (EDA) ✅

📘 MCQ Answer Key

1.​ C – Understand and explore data
2.​ B – Data understanding
3.​ B – Data quality check
4.​ C – Univariate analysis
5.​ C – Histogram
6.​ C – Bivariate analysis
7.​ B – Bivariate analysis
8.​ B – Outlier detection
9.​ B – Detect outliers
10.​C – Drawing insights and conclusions

👍
Alright ​
Here’s 3.2 Appropriate features / variables are identified for the analysis — clear
explanation first, then exam-ready MCQs, same pattern as before.

3.2 Appropriate features / variables are identified for the

analysis – Explanation
Identifying appropriate features (variables) means selecting the most relevant variables from
the dataset that help answer the analysis objective or solve the business problem.

Not all variables are useful. Some may be irrelevant, redundant, or noisy, which can
negatively affect analysis and model performance.

🔹 What this step involves:

 ●​ Understanding the target variable (what you want to predict or analyze)
●​ Identifying input (independent) variables that influence the target
●​ Removing irrelevant or duplicate variables
●​ Considering domain knowledge and EDA insights
●​ Checking correlation and importance of variables

🔹 Why this is important:

●​ Improves accuracy and reliability of analysis
●​ Reduces overfitting
●​ Makes models simpler and interpretable
●​ Saves computational time

🔹 Example:
For predicting student performance:

●​ Relevant features: study hours, attendance, previous scores

●​ Irrelevant features: student ID, registration number

MCQs: 3.2 Feature / Variable Identification

1.​ Identifying appropriate features means selecting variables that are:​
A. Random​
B. Relevant to the analysis objective​
C. Always numerical​
D. Duplicated
2.​ The variable that is predicted or analyzed is called the:​
A. Independent variable​
B. Input variable​
C. Target variable​
D. Noise variable
3.​ Which variables are usually removed during feature identification?​
A. Important variables​
B. Relevant variables​
C. Irrelevant or redundant variables​
D. Target variables
4.​ Correlation analysis helps in:​
A. Data collection​
B. Identifying relationships between variables​
C. Data storage​
D. Model deployment
 5.​ Domain knowledge is useful in feature selection because it:​
A. Replaces data​
B. Helps identify meaningful variables​
C. Increases dataset size​
D. Removes the target variable
6.​ Which feature is LEAST useful for predicting house prices?​
A. Location​
B. Size of house​
C. Number of rooms​
D. House ID number
7.​ Selecting too many irrelevant features may lead to:​
A. Better performance​
B. Overfitting​
C. Perfect accuracy​
D. No effect
8.​ Feature identification is mainly part of:​
A. Data Collection​
B. Exploratory Data Analysis​
C. Model Deployment​
D. Monitoring
9.​ Independent variables are also known as:​
A. Output variables​
B. Predictor variables​
C. Noise variables​
D. Target variables
10.​Identifying appropriate features helps to:​
A. Increase noise​
B. Reduce model interpretability​
C. Improve analysis and model performance​
D. Increase missing values

✅
Here are the answers for 3.2 Appropriate features / variables are identified for the analysis

📘 MCQ Answer Key

1.​ B – Relevant to the analysis objective
2.​ C – Target variable
3.​ C – Irrelevant or redundant variables
4.​ B – Identifying relationships between variables
5.​ B – Helps identify meaningful variables
 6.​ D – House ID number
7.​ B – Overfitting
8.​ B – Exploratory Data Analysis
9.​ B – Predictor variables
10.​C – Improve analysis and model performance

Nice, continuing smoothly 👍 ​

Here’s 3.3 Statistical characteristics of the variables are analyzed — clear explanation
first, then exam-ready MCQs, same pattern as before.

3.3 Statistical characteristics of the variables are

analyzed – Explanation
Analyzing statistical characteristics means summarizing and understanding variables using
numerical measures. This step helps describe the distribution, central tendency, spread,
and shape of the data.

It is a core part of Exploratory Data Analysis (EDA) and helps detect patterns, anomalies,
and data quality issues.

🔹 Key statistical characteristics analyzed:

1. Measures of Central Tendency

●​ Mean – average value

●​ Median – middle value
●​ Mode – most frequent value

2. Measures of Dispersion

●​ Range – difference between max and min

●​ Variance – spread of data
●​ Standard deviation – average deviation from mean

3. Shape of Distribution

●​ Skewness – symmetry of data

●​ Kurtosis – peakedness or flatness
4. Other statistics

●​ Minimum and maximum values

●​ Quartiles and percentiles

🔹 Why this is important:

●​ Helps understand data distribution
●​ Identifies outliers and skewness
●​ Guides data transformation decisions
●​ Supports better model selection

🔹 Example:
Analyzing exam scores using mean, median, standard deviation, and skewness to understand
student performance distribution.

MCQs: 3.3 Statistical Characteristics

1.​ Statistical characteristics are mainly used to:​
A. Store data​
B. Summarize and understand data​
C. Deploy models​
D. Collect data
2.​ Which measure represents the average value of a dataset?​
A. Median​
B. Mode​
C. Mean​
D. Range
3.​ The median of a dataset is:​
A. The most frequent value​
B. The middle value​
C. The average value​
D. The maximum value
4.​ Which measure describes the spread of data?​
A. Mean​
B. Median​
C. Standard deviation​
D. Mode
 5.​ Variance is defined as:​
A. Square root of standard deviation​
B. Average of squared deviations from the mean​
C. Difference between max and min​
D. Middle value of data
6.​ Skewness indicates:​
A. Data accuracy​
B. Data size​
C. Symmetry of distribution​
D. Data format
7.​ Positive skewness means:​
A. Left-tailed distribution​
B. Symmetric distribution​
C. Right-tailed distribution​
D. Uniform distribution
8.​ Kurtosis measures the:​
A. Central value​
B. Spread​
C. Peakedness of distribution​
D. Data size
9.​ Which statistic helps identify outliers?​
A. Mean only​
B. Standard deviation​
C. Mode​
D. Data type
10.​Analyzing statistical characteristics is part of:​
A. Data Collection​
B. Exploratory Data Analysis​
C. Model Deployment​
D. Monitoring

Here are the answers for 3.3 Statistical characteristics of the variables are analyzed ✅

📘 MCQ Answer Key

1.​ B – Summarize and understand data
2.​ C – Mean
3.​ B – The middle value
4.​ C – Standard deviation
 5.​ B – Average of squared deviations from the mean
6.​ C – Symmetry of distribution
7.​ C – Right-tailed distribution
8.​ C – Peakedness of distribution
9.​ B – Standard deviation
10.​B – Exploratory Data Analysis

Great 👍 ​
Here’s 3.4 and 3.5 done in the same exam-ready format: explanation first, then MCQs for
both.

3.4 Dataset is parsed by cleaning, treating missing values

& outliers and transforming as required – Explanation
Parsing a dataset means preparing and refining the raw data so it becomes suitable for
deeper analysis.

This step combines multiple preprocessing activities to improve data quality and reliability.

🔹 Key activities involved:

●​ Cleaning
○​ Removing duplicates
○​ Fixing inconsistent or invalid values
●​ Treating missing values
○​ Deletion or imputation (mean, median, mode)
●​ Handling outliers
○​ Detection using IQR or Z-score
○​ Removal, capping, or transformation
●​ Transforming data
○​ Scaling and normalization
○​ Encoding categorical variables
○​ Log or power transformation

🔹 Why this step is important:

●​ Ensures accurate statistical analysis
●​ Prevents model bias and errors
●​ Improves performance of algorithms
3.5 Bi-variate and multivariate analysis are performed to
find associations between variables – Explanation
This step focuses on analyzing relationships between two or more variables to understand
how they influence each other.

🔹 Types of analysis:
Bivariate analysis

●​ Examines relationship between two variables

●​ Techniques:
○​ Correlation
○​ Scatter plots
○​ Crosstabs

Multivariate analysis

●​ Examines relationships among three or more variables

●​ Techniques:
○​ Multiple regression
○​ Heatmaps
○​ Pair plots
○​ PCA

🔹 Why this is important:

●​ Helps identify associations and dependencies
●​ Supports feature selection
●​ Improves model design and interpretation

MCQs: 3.4 Dataset Parsing

1.​ Dataset parsing mainly involves:​
A. Data collection​
B. Data cleaning and transformation​
C. Model deployment​
D. Visualization only
 2.​ Removing duplicate records is part of:​
A. Data collection​
B. Data cleaning​
C. Feature selection​
D. Sampling
3.​ Missing values are commonly treated using:​
A. Encoding​
B. Imputation​
C. Indexing​
D. Visualization
4.​ Which method is used to detect outliers?​
A. Normalization​
B. Z-score​
C. Encoding​
D. Aggregation
5.​ Encoding categorical variables is an example of:​
A. Data deletion​
B. Data transformation​
C. Data visualization​
D. Data collection

MCQs: 3.5 Bi-variate & Multivariate Analysis

6.​ Bivariate analysis studies the relationship between:​
A. One variable​
B. Two variables​
C. Three variables​
D. All variables
7.​ Scatter plots are mainly used in:​
A. Univariate analysis​
B. Bivariate analysis​
C. Multivariate analysis only​
D. Data cleaning
8.​ Correlation measures:​
A. Data accuracy​
B. Strength of relationship between variables​
C. Missing values​
D. Data size
 9.​ Multivariate analysis involves:​
A. One variable only​
B. Two variables only​
C. Three or more variables​
D. Target variable only
10.​Which technique is used in multivariate analysis?​
A. Histogram​
B. Box plot​
C. Multiple regression​
D. Mode calculation

Here are the answers for 3.4 Dataset parsing and 3.5 Bi-variate & multivariate analysis ✅

📘 MCQ Answer Key

3.4 Dataset is parsed (cleaning, missing values, outliers, transformation)

1.​ B – Data cleaning and transformation

2.​ B – Data cleaning
3.​ B – Imputation
4.​ B – Z-score
5.​ B – Data transformation

3.5 Bi-variate & Multivariate Analysis

6.​ B – Two variables

7.​ B – Bivariate analysis
8.​ B – Strength of relationship between variables
9.​ C – Three or more variables
10.​C – Multiple regression
4.1 Visualization of variables and relationships –
Explanation
Data visualization is the process of representing data graphically to make patterns, trends,
distributions, and relationships easier to understand.

Different data types require different visualization methods to correctly represent information.

🔹 Visualization methods by data type:

1. Numerical data

●​ Histogram → distribution of a single variable

●​ Box plot → spread and outliers
●​ Line chart → trends over time

2. Categorical data

●​ Bar chart → category comparison

●​ Pie chart → proportion of categories

3. Relationships between variables

●​ Scatter plot → relationship between two numerical variables

●​ Heatmap → correlation between multiple variables
●​ Pair plot → multivariate relationships

4. Time-series data

●​ Line plot → trends over time

🔹 Why visualization is important:

●​ Makes patterns and relationships visible
●​ Helps detect outliers and anomalies
●​ Supports data-driven decision making
●​ Improves communication of insights
MCQs: 4.1 Data Visualization
1.​ The main purpose of data visualization is to:​
A. Store data​
B. Make patterns and relationships easy to understand​
C. Delete data​
D. Encode data
2.​ Which visualization is BEST for showing the distribution of a numerical variable?​
A. Bar chart​
B. Histogram​
C. Pie chart​
D. Line chart
3.​ Which chart is most suitable for categorical data comparison?​
A. Scatter plot​
B. Bar chart​
C. Histogram​
D. Box plot
4.​ A scatter plot is mainly used to:​
A. Show category proportions​
B. Show trends over time​
C. Show relationship between two numerical variables​
D. Detect missing values
5.​ Which visualization helps identify outliers?​
A. Pie chart​
B. Line chart​
C. Box plot​
D. Bar chart
6.​ Heatmaps are commonly used to visualize:​
A. Text data​
B. Missing values only​
C. Correlation between variables​
D. Categorical distributions
7.​ Line charts are best suited for:​
A. Categorical comparison​
B. Time-series data​
C. Distribution analysis​
D. Feature encoding
8.​ Which visualization shows proportions of categories?​
A. Histogram​
B. Scatter plot​
C. Pie chart​
D. Box plot
 9.​ Pair plots are useful for:​
A. Univariate analysis only​
B. Bivariate analysis only​
C. Multivariate analysis​
D. Data storage
10.​Choosing an inappropriate visualization may result in:​
A. Clear insights​
B. Misleading interpretation​
C. Better accuracy​
D. Improved performance

✅
Here are the answers for 4.1 Variables and relationships are visualized using various
visualization methods

📘 MCQ Answer Key

1.​ B – Make patterns and relationships easy to understand
2.​ B – Histogram
3.​ B – Bar chart
4.​ C – Show relationship between two numerical variables
5.​ C – Box plot
6.​ C – Correlation between variables
7.​ B – Time-series data
8.​ C – Pie chart
9.​ C – Multivariate analysis
10.​B – Misleading interpretation

4.2 Python visualization libraries or BI tools are used –

Explanation
Data visualization tools are used to create charts and graphs that help understand data
visually.

Common tools:

●​ Python libraries
○​ Matplotlib → basic, static plots
○​ Plotly → interactive and dynamic plots
 ●​ Business Intelligence (BI) tools
○​ Power BI
○​ Tableau

Why these tools are used:

●​ Convert data into visual insights

●​ Support interactive exploration
●​ Create professional dashboards and reports

4.3 Plots are analyzed to identify important patterns –

Explanation
Once plots are created, they are interpreted to extract insights such as:

●​ Trends
●​ Correlations
●​ Clusters
●​ Outliers
●​ Seasonal behavior

Examples:

●​ Line plot → upward or downward trends

●​ Scatter plot → positive or negative correlation
●​ Box plot → outliers and spread

This step transforms visuals into actionable insights.

4.4 Reports are generated using Power BI or Tableau –

Explanation
Reports combine multiple visualizations, filters, and summaries into a structured format.

Features of BI reports:

●​ Dashboards with charts and KPIs

●​ Interactive filters and slicers
●​ Drill-down capabilities
Purpose:

●​ Present insights clearly

●​ Support decision-making
●​ Communicate results effectively

4.5 Reports are shared with stakeholders – Explanation

After report creation, insights are shared with stakeholders such as managers, clients, or
executives.

Sharing methods:

●​ Power BI Service / Tableau Server

●​ PDF or image export
●​ Online dashboards
●​ Email or cloud sharing

Importance:

●​ Enables data-driven decisions

●​ Aligns technical insights with business goals
●​ Improves communication and transparency

MCQs: 4.2 Visualization Tools

1.​ Which Python library is commonly used for basic plotting?​
A. NumPy​
B. Matplotlib​
C. Pandas​
D. Scikit-learn
2.​ Which tool provides interactive visualizations?​
A. Matplotlib only​
B. Plotly​
C. Excel only​
D. SQL
 3.​ Power BI and Tableau are examples of:​
A. Programming languages​
B. Databases​
C. Business Intelligence tools​
D. Operating systems

MCQs: 4.3 Plot Analysis

4.​ Scatter plots are mainly analyzed to identify:​
A. Trends over time​
B. Correlation between variables​
C. Category proportions​
D. Missing values
5.​ Box plots help identify:​
A. Mean values​
B. Outliers​
C. Categories​
D. Trends
6.​ An upward trend in a line chart indicates:​
A. Decreasing values​
B. No change​
C. Increasing values​
D. Random behavior

MCQs: 4.4 Report Generation

7.​ Dashboards are mainly used to:​
A. Store raw data​
B. Present insights visually​
C. Clean data​
D. Train models
8.​ Which feature allows users to explore data at different levels?​
A. Encoding​
B. Drill-down​
C. Sampling​
D. Normalization
MCQs: 4.5 Report Sharing
9.​ Reports are shared with stakeholders mainly to:​
A. Increase data size​
B. Enable decision making​
C. Clean data​
D. Store data
10.​Which is a common method to share BI reports?​
A. Compiler​
B. Power BI Service​
C. Text editor​
D. Command line