BDA-DSC-4 (Maj)-401

Data Collection & Methodology

UNIT 1 — Complete Study Notes
BSc Data Analytics | 4th Semester
Theory External: 90 Marks | Duration: 3 Hours

BDA-DSC-4 | Unit 1 Study Notes | Page 1

 TOPIC 1: What is Data?

📌 Data: Raw, unprocessed facts and figures that, on their own, have no meaning. Data becomes
information when it is processed, organized, and given context.

Think of it this way: Data is like raw ingredients. Information is the finished meal.

Data vs. Information — Key Difference

✏️Example: Raw data: 25, 30, 22, 28. → Information: 'Average daily temperatures in Delhi over
4 days in June 2024.'
✏️Example: Raw data: 9845673210, 18, M, 560001 → Information: Customer record of an 18-
year-old male in Bangalore.

Why Data Matters in Analytics

In today's digital world, data is called the 'new oil.' Organizations that collect and analyze data correctly
gain massive advantages over those that rely purely on intuition. Data analytics transforms raw data
into actionable insights — helping leaders make smarter decisions.

📝 EXAM TIP: In an exam, always start your answer with a clear definition. Then add an example.
Examiners reward structured answers. Define → Explain → Example = guaranteed marks.

TOPIC 2: Importance of Data-Driven Decision Making

Data-driven decision making (DDDM) means making choices based on actual data and analysis rather
than intuition or guesswork. In the modern business world, organizations that use data outperform
those that don't.

Why is it Important? (Write these as separate points in exams)

• Reduces Risk — Decisions based on evidence are more reliable. Instead of guessing what
customers want, you analyze their buying behavior.
• Improves Efficiency — Data reveals bottlenecks and inefficiencies in operations. Companies can
then optimize resources.
• Better Customer Understanding — Analyzing purchase history and preferences helps businesses
tailor their products/services to customer needs.
• Competitive Advantage — Companies using data analytics can predict market trends before
competitors, giving them an edge.
• Identifies New Opportunities — Data analysis can reveal untapped customer segments or
product demands that weren't obvious before.
• Accountability & Transparency — Decisions backed by data are easier to justify to stakeholders
and shareholders.

BDA-DSC-4 | Unit 1 Study Notes | Page 2

 ✏️Example: Netflix uses viewer data (what you watch, when you pause, what you rewatch) to
decide which original shows to produce. This data-driven approach led to hits like Stranger Things.
Without data, they'd just be guessing.
✏️Example: Amazon uses purchase and browsing data to recommend products — 35% of their
revenue comes from these data-driven recommendations.

📝 EXAM TIP: For a long-answer question on DDDM, write 5–6 points with examples. That alone
can get you 8–10 marks easily.

TOPIC 3: Types of Data

This is one of the MOST important topics in Unit 1. Multiple questions can come from here.
Learn all sub-categories thoroughly.

◆ A) Primary vs. Secondary Data

Feature Primary Data Secondary Data

Definition Data collected fresh by the Data already collected by someone

researcher for a specific purpose else for a different purpose
Source Surveys, interviews, experiments, Government reports, journals,
observations census data, published research
Cost Expensive and time-consuming Cheaper and faster to access
Relevance Highly specific to your research May not perfectly match your need
need
Accuracy Researcher controls the quality Quality depends on the original
collector
Freshness Current and up-to-date May be outdated
Example Survey asking 200 students their Using RBI's published data on bank
study habits interest rates

📝 EXAM TIP: Always write a comparison table when asked to 'compare' or 'distinguish.' It shows
structure and earns more marks than a paragraph explanation.

◆ B) Structured, Semi-structured, and Unstructured Data

1. Structured Data

BDA-DSC-4 | Unit 1 Study Notes | Page 3

Structured data is highly organized data that fits neatly into rows and columns — like a spreadsheet or
database table. It is easy to store, search, and analyze using standard tools.
• Format — Tables, spreadsheets, relational databases (SQL)
• Tools — MS Excel, MySQL, Oracle, Google Sheets
• Searchability — Very easy — you can filter, sort, and run queries
• Volume — Makes up only about 20% of all data in the world
✏️Example: A student database with columns: Roll No, Name, Marks, Attendance, Grade — this
is structured data.

2. Semi-structured Data
Semi-structured data has SOME organization but does NOT fit perfectly into a rigid table. It uses tags,
markers, or keys to separate elements, making it partially searchable.
• Format — JSON, XML, HTML, CSV files, email (has fields like To/From/Subject but the body is
free text)
• Tags — Uses self-describing tags or key-value pairs
• Flexibility — More flexible than structured but harder to analyze than fully structured
✏️Example: A JSON file: { 'name': 'Rahul', 'age': 20, 'courses': ['Analytics', 'Statistics'] } — some
structure, but the 'courses' field is a list, not a fixed column.
✏️Example: An email has structured fields (To, From, Date, Subject) but the body text is
unstructured.

3. Unstructured Data
Unstructured data has NO fixed format or organization. It is the most common type of data in the world
and the hardest to analyze without advanced tools like AI or NLP.
• Format — Videos, images, audio files, social media posts, PDFs, Word documents
• Volume — Makes up approximately 80% of all data generated globally
• Analysis — Requires special tools — Natural Language Processing (NLP), Computer Vision,
Machine Learning
✏️Example: A WhatsApp group chat, a YouTube video, a customer review on Amazon, a doctor's
handwritten notes — all are unstructured data.

Feature Structured Semi-structured Unstructured

Format Fixed rows & columns Partial tags/markers No fixed format

Storage Relational DB (SQL) NoSQL, JSON, XML Data lakes, cloud storage
Searchability Very Easy Moderate Difficult
% of world data ~20% ~Smaller portion ~80%
Example Excel sheet, SQL JSON, XML, Email Videos, images, tweets
table

📝 EXAM TIP: Writing a 3-way comparison table for structured/semi-structured/unstructured can

earn you 6–8 marks on its own. Practice writing it from memory.

BDA-DSC-4 | Unit 1 Study Notes | Page 4

 ◆ C) Cross-sectional, Longitudinal, Time Series, and Panel Data

1. Cross-sectional Data
Data collected from many different subjects at a single point in time. Think of it as a photograph — it
captures one moment across many people/entities.
• Focus — Multiple subjects, ONE time point
• Use — Comparing groups, understanding patterns at a specific moment
✏️Example: A survey of 500 BSc students' exam scores in April 2025. All data collected at one
time, from many students.
✏️Example: Census data collected in 2011 — millions of people, one point in time.

2. Longitudinal Data
Data collected from the SAME subjects over a long period of time. It tracks how things change over
time for specific individuals or entities.
• Focus — Same subjects, MULTIPLE time points over a long period
• Use — Tracking growth, development, or change over time
✏️Example: Tracking the health of 100 patients every year for 20 years to study the effects of a
drug.
✏️Example: A company tracking 50 employees' productivity every quarter for 5 years.

3. Time Series Data

Data points collected at regular, fixed time intervals for ONE variable. The focus is on how a single
thing changes over time at consistent intervals.
• Focus — ONE variable, collected at REGULAR intervals over time
• Use — Forecasting, trend analysis, seasonal patterns
✏️Example: Daily stock prices of Infosys from January to December 2024.
✏️Example: Monthly rainfall in Mumbai for the past 10 years. Weekly sales figures of a
supermarket.

4. Panel Data (Cross-sectional + Longitudinal)

Panel data combines both approaches — it tracks MULTIPLE subjects over MULTIPLE time periods.
Also called longitudinal cross-sectional data. It is the most comprehensive data type.
• Focus — Many subjects, many time periods
• Use — Complex economic and social research, policy analysis
✏️Example: GDP of 10 countries recorded every year for 30 years — multiple countries
(subjects), multiple years (time points).
✏️Example: Test scores of 200 students measured every semester over 3 years.

BDA-DSC-4 | Unit 1 Study Notes | Page 5

 Data Type Subjects Time Points Best For Example

Cross-sectional Many One Comparing groups at a Survey of 500

moment students' marks in
April 2025
Longitudinal Same set Many (long period) Tracking change over Health of 100
time patients over 20
years
Time Series One variable Regular intervals Forecasting, trends Daily stock prices for
1 year
Panel Many Many Comprehensive analysis GDP of 10 countries,
30 years

🧠 Memory Trick:
Cross-sectional = Many people, ONE moment (like a class photo) | Time Series = ONE thing,
many moments | Panel = Many things, many moments

TOPIC 4: Sources of Data

Understanding WHERE data comes from is as important as knowing what types it is. Different sources
suit different research goals.

1. Surveys
A survey is a set of standardized questions (questionnaire) given to a group of people to gather
opinions, preferences, or facts.
• Types — Online surveys (Google Forms), telephonic, in-person, postal
• Best for — Collecting attitudes, opinions, preferences from large groups
• Advantage — Can reach many respondents quickly and cheaply
• Disadvantage — Respondents may not answer honestly; response rates can be low
✏️Example: A company sends a customer satisfaction survey via email after every purchase. This
is a survey as a data source.

2. Interviews
An interview is a direct conversation between a researcher and a participant to gather in-depth
information.
• Structured Interview — Fixed set of questions asked in a specific order — like a written survey but
spoken
• Semi-structured Interview — Core questions prepared but allows follow-up and flexibility
• Unstructured Interview — Open conversation with no fixed questions — most flexible, most time-
consuming
• Best for — Collecting deep insights, opinions, personal experiences

BDA-DSC-4 | Unit 1 Study Notes | Page 6

 ✏️Example: A researcher interviewing doctors about challenges in rural healthcare — they'd use a
semi-structured interview to allow for follow-up.

3. Observations
The researcher watches subjects directly and records their behavior WITHOUT interfering or asking
questions.
• Participant Observation — Researcher joins the group being studied
• Non-participant Observation — Researcher watches from the outside without being part of the
group
• Best for — Understanding natural behavior, which people may not accurately self-report
✏️Example: Observing how shoppers move through a supermarket to understand which product
sections attract most attention — used by retail chains to decide shelf placement.

4. Experiments
A controlled research method where one or more variables are manipulated to observe the effect on an
outcome. Establishes cause-and-effect relationships.
• A/B Testing — Showing two different versions of a product/website to two groups to see which
performs better
• Controlled Experiment — One group gets the treatment; another (control group) does not
• Best for — Testing hypotheses, proving causation (not just correlation)
✏️Example: Flipkart tests two versions of their homepage — Version A with a red 'Buy Now'
button and Version B with a blue one — to see which gets more clicks. This is A/B testing (an
experiment).

5. Administrative and Transactional Data

This data is automatically generated as a by-product of running an organization's daily operations. No
special effort is needed to collect it — it exists because transactions happen.
• Administrative Data — Hospital patient records, school attendance registers, government tax
records
• Transactional Data — Bank transaction logs, retail sales receipts, online order histories
• Advantage — Huge volume, collected automatically, very low cost
✏️Example: Every time you use your UPI to pay, a transaction record is generated. Banks use
this transactional data to detect fraud and suggest financial products.

6. Web Data (Digital Sources)

Data generated through internet activity — the fastest growing source of data today.
• Social Media Data — Posts, likes, shares, comments, hashtags on platforms like Instagram,
Twitter/X, LinkedIn
• E-commerce Data — Purchase history, product reviews, browsing behavior, cart abandonment
data
• Sensor / IoT Data — Data from smart devices — fitness trackers (Fitbit), smart thermostats, traffic
sensors, weather stations

BDA-DSC-4 | Unit 1 Study Notes | Page 7

• Web Scraping — Automatically extracting data from websites using programs/scripts
✏️Example: Zomato uses web and app data (orders, reviews, location data, delivery time) to
recommend restaurants and predict demand in different areas.
✏️Example: A smart fitness tracker records your heart rate every second — this is IoT/sensor
data.

📝 EXAM TIP: If asked to write about 'sources of data,' write at least 5 sources with a brief
explanation and one example each. Aim for 2–3 sentences per source. That's an easy 10-mark
answer.

TOPIC 5: Data Collection Objectives and Planning

Before any data is collected, a researcher must plan carefully. Rushing into collection without clear
objectives wastes time, money, and effort — and produces unreliable results.

Step 1 — Define the Research Objective

The most critical step. You must clearly state WHAT question you are trying to answer. A vague
objective leads to vague data.
• Ask yourself — What specific problem am I solving? What decision will this data support?
✏️Example: Bad objective: 'Study students.'
✏️Example: Good objective: 'Identify the top 3 factors affecting exam performance of BSc Data
Analytics students in Karnataka in 2025.'

Step 2 — Define the Scope

Scope defines the BOUNDARIES of your study — who will be studied, what geographic area, what
time period, and what aspects.
• Population — Who is the target group? (All BSc students? Only 4th-semester students?)
• Geography — Which location? (One college? All of Karnataka? Pan-India?)
• Time Period — What dates? (Academic year 2024–25? Last 5 years?)
• Variables — What will you measure? (Marks? Attendance? Study hours?)
✏️Example: Scope: 'BSc Data Analytics 4th-semester students across 10 colleges in Bengaluru,
during the academic year 2024–25, measuring attendance, study hours, and final exam scores.'

Step 3 — Choose the Appropriate Data Collection Method

The method must match your objective, scope, budget, and timeline. There is no universal best method
— it depends on what you need.
If your goal is... Best Method

Gather opinions from a large group Survey / Questionnaire

Understand deep personal experiences Interview

BDA-DSC-4 | Unit 1 Study Notes | Page 8

 Study natural behavior Observation
Prove cause and effect Experiment / A/B Test
Analyze existing records Administrative / Transactional data
Track online trends Web scraping / Social media data

📝 EXAM TIP: Questions like 'What factors should be considered before collecting data?' or
'Explain the planning process for data collection' come regularly. Use these 3 steps and expand
each with examples for a full 15-mark answer.

TOPIC 6: Ethical Issues in Data Collection

Ethics in data collection means ensuring that the process of gathering data is fair, honest, and
respectful of people's rights. Unethical data collection can harm individuals, damage trust, and even
lead to legal consequences.

Key Ethical Principles — Write Each with Explanation

📌 1. Informed Consent: Participants must be fully informed about what the study involves, why
their data is needed, and how it will be used — BEFORE they agree to participate. Consent must
be voluntary, not coerced.
✏️Example: Before conducting a survey in a college, students must be told: 'This study is about
stress levels among students. Your responses are voluntary and will be used only for academic
research.'

📌 2. Privacy and Confidentiality: Personal information shared by participants must NOT be

revealed to unauthorized parties. Data should only be used for the purpose it was collected for.
✏️Example: A hospital collects patient data for a study. The data must not be shared with
insurance companies or third parties without patient permission.

📌 3. Anonymity: Participants' identities should be protected. Their responses should not be

traceable back to them individually. Anonymity is stronger than confidentiality — even the
researcher shouldn't know who said what.
✏️Example: In a survey about workplace harassment, employee names should not be recorded
so respondents can answer honestly without fear.

📌 4. Avoiding Deception: Researchers must not mislead participants about the true purpose of
the study. If deception is necessary (e.g., in some psychology experiments), participants must be
fully debriefed afterward.
✏️Example: You cannot tell participants 'This is a study about memory' when it is actually testing
their reaction to stress. That is deception.

BDA-DSC-4 | Unit 1 Study Notes | Page 9

📌 5. Data Security: All collected data must be stored securely to prevent unauthorized access,
theft, or breaches. This includes encryption, secure servers, and access controls.
✏️Example: Survey responses stored on a password-protected, encrypted cloud server — not on
a personal email inbox that can be easily hacked.

📌 6. Right to Withdraw: Participants must have the freedom to withdraw from the study at any
point, without facing any negative consequences. Participation is always voluntary.
✏️Example: During a longitudinal health study, a participant can decide in year 3 to stop sharing
their data, and the researcher must respect that.

📌 7. Avoiding Harm (Non-maleficence): The research process must not cause physical,
psychological, social, or financial harm to participants. If harm is a risk, it must be minimized and
participants warned.
✏️Example: A study that asks trauma survivors to relive painful experiences without providing
counseling support could cause psychological harm — this is unethical.

📌 8. Accuracy and Honesty: Data must be reported truthfully and completely. Researchers must
not manipulate, cherry-pick, or falsify data to support a desired conclusion. This is called data
integrity.
✏️Example: A pharmaceutical company hiding negative trial results about a drug and only
publishing positive results is a major ethical violation.

Ethical Issue What It Means Why It Matters

Informed Consent Participants agree after being Respects autonomy and prevents
fully informed exploitation
Privacy & Data not shared without Prevents misuse of personal
Confidentiality permission information
Anonymity Identities protected Encourages honest responses
Avoiding Deception No misleading about study Maintains trust in research
purpose
Data Security Safe storage of collected data Prevents breaches and misuse
Right to Withdraw Can exit study anytime Ensures voluntary participation
Avoiding Harm No physical or mental damage Protects participant wellbeing
Accuracy & Honesty Data reported truthfully Ensures research validity

📝 EXAM TIP: A 10-mark question on ethics = write 5–6 points, each with 2–3 sentences and an
example. Don't just list them — explain WHY each one matters. That difference gets you distinction
marks.

QUICK REVISION SUMMARY — Unit 1

BDA-DSC-4 | Unit 1 Study Notes | Page 10

 Topic What to Remember for Exams

Data Definition Raw unprocessed facts → becomes Information when processed and
given context
Data-Driven Decisions 6 benefits: Reduces risk, improves efficiency, customer understanding,
competitive advantage, identifies opportunities, accountability
Primary Data Collected fresh by researcher | Surveys, interviews, experiments |
Expensive but specific
Secondary Data Pre-existing data | Reports, census, journals | Cheaper but may not
perfectly fit
Structured Data Rows & columns, SQL databases, Excel | Easy to analyze | 20% of
world data
Semi-structured Partial tags/keys | JSON, XML, Email | Moderate difficulty to analyze
Unstructured No format | Videos, images, social media posts | 80% of world data |
Hardest to analyze
Cross-sectional Many subjects, ONE time point | Snapshot | Example: Annual census
Longitudinal SAME subjects, many time points | Tracks change | Example: Patient
health over 20 years
Time Series ONE variable, regular intervals | Forecasting | Example: Monthly stock
prices
Panel Data Many subjects + many times | Most comprehensive | Example: GDP of
10 countries for 30 years
Data Sources Surveys, Interviews, Observations, Experiments,
Administrative/Transactional, Web/IoT
Collection Planning 3 steps: Define Objective → Define Scope → Choose Method
Ethics (8 points) Consent, Privacy, Anonymity, No Deception, Data Security, Right to
Withdraw, No Harm, Accuracy

◆ Exam Strategy for Unit 1

• For 'Define and Explain' questions — Always: Definition → Explanation → 2 Examples →

Advantages/Disadvantages if applicable
• For 'Compare' questions — Always draw a table — it shows structure and saves time
• For 'List and Explain' questions — Number your points clearly. Write 2–3 sentences per point
minimum
• For Ethics questions — Write each ethical principle as a numbered heading, then explain, then
give example
• Word volume matters — For a 15-mark answer aim for 1.5 to 2 pages. Practice writing in full
sentences, not bullet fragments

📝 EXAM TIP: You have 4 questions to attempt (1 per unit) plus 1 compulsory short-answer
question. For Unit 1, the most likely long-answer topics are: Types of Data, Sources of Data,
Ethical Issues in Data Collection, and Data-Driven Decision Making. Prepare all four in detail!

BDA-DSC-4 | Unit 1 Study Notes | Page 11

BDA-DSC-4 | Unit 1 Study Notes | Page 12