Introduction

to Statistics
HISTORY, CONCEPTS, AND APPLICATIONS IN DATA
SCIENCE & ANALYTICS
Wath is expected in Interviews?
•They test core statistics: proportions, hypothesis
testing, t-test.
•No ML or coding needed → pure statistical
reasoning.
•They see if you can connect formulas → business
meaning.
“Statistical thinking will be one day as
necessary for efficient citizenship as the
ability to read and write”

H G Wells
History of Statistics

 - Originated from the Latin word 'Status' meaning

state
 - Early use: Governments collecting data on
population, taxation
 - 17th century: John Graunt studied mortality rates
 - 18th century: Statistics developed as a field of
mathematics
 - Modern era: Foundation of Data Science and
Analytics
Definition of Statistics

 Statistics is the science of collecting, organizing,

analyzing, and interpreting data.

 Example: Calculating the average marks of a

class from student test scores.
Statistics may be defined as science that is employed
to
 Collect the data
 Present and organize the data in a systematic
manner
 Analyse the data
 Infer about the data
 Take decision from the data.

In other words, Statistics can also be defined as

numerical data with a view to analyse it.
Use Cases in Data Science
& Analytics
 - Customer behavior analysis (e.g., e-commerce
recommendation systems)
 - Fraud detection in banking
 - Healthcare: Predicting disease risk from patient
data
 - Sports analytics: Player performance evaluation
 - Business forecasting: Sales trends, demand
prediction
 Types of Variable

Qualitative (Categorical): Quantitative(Numerical):

express a qualitative attribute measured in terms of numbers
such as hair color, eye color, such as height, weight,
religion. number of people.

Nominal: Ordinal:
Discrete: Continuous: not
No ordering is Ordering is possible
countable and countable and
possible such as such as health,
have a finite have an infinite
hair color, eye which can take
number of number of
color, religion. values such as poor,
possibilities such possibilities such
reasonable, good,
as number of as height
or excellent.
people

INTERVAL: ratio of values of

RATIO: ratio of values of
variable do not have any
variable have meaning and it
meaning and it does not have
have an inherently defined zero
an inherently defined zero value
value such as length
such as temperature
Discrete Variables

Definition:
•Can take only specific, countable values (usually whole
numbers).
•Comes from counting things.
•No decimals/fractions possible.

Examples:
[Link] of children in a family → 0, 1, 2, 3 … (not 2.5).
[Link] of books on a shelf → 10, 11, 12 …
[Link] of students present in class → 25, 26 …
[Link] of goals scored in a football match → 0, 1, 2 …
[Link] outcome → 1 to 6.

Key point: Discrete = whole numbers, you count them.

Continuous Variables

Definition:
•Can take any value within a range (infinite possibilities).
•Comes from measuring things.
•Can have decimals/fractions.

Examples:
[Link] of students → 160.2 cm, 160.25 cm …
[Link] of a person → 55.5 kg, 55.52 kg …
[Link] taken to finish a race → 12.1 sec, 12.15 sec …
[Link] → 28.3°C, 28.33°C …
[Link] between two cities → 250.4 km, 250.45 km …

Key point: Continuous = decimals allowed, you measure the

Simple Analogy

•Discrete variable = Counting chocolates 🍫 in a box (you can’t

have 2.5 chocolates).

•Continuous variable = Measuring milk 🥛 in a glass (200 ml, 200.5

ml, 200.55 ml …).
Interval Variables

Definition:

•Numerical values with equal intervals (difference makes sense).

•But they don’t have a true zero (zero does not mean “nothing”).

•So, you can add and subtract, but you cannot multiply or divide
meaningfully.
Examples of Interval Variables:
[Link] in Celsius or Fahrenheit
1. Difference between 30°C and 20°C = 10°C →
meaningful.
2. But 0°C ≠ “no temperature” (it just means freezing point).
3. You cannot say “40°C is twice as hot as 20°C.”
[Link] years
1. Difference between 2000 and 1990 = 10 years →
meaningful.
2. But “Year 0” doesn’t mean no time existed.
[Link] Scores
1. Difference between IQ 120 and IQ 100 = 20 →
meaningful.
2. But 0 IQ does not mean “no intelligence.”

✅ Key Idea: Interval = equal gaps matter, but zero is arbitrary.

Ratio Variables

Definition:

•Numerical values with equal intervals.

•Have a true zero (zero means “nothing”).

•You can do all math: add, subtract, multiply, divide.

Examples of Ratio Variables:
[Link] & Weight
1. 0 kg = no weight.
2. 60 kg is twice as heavy as 30 kg.
[Link] taken to complete a task
1. 0 sec = no time passed.
2. 20 sec is twice as long as 10 sec.
[Link]
1. 0 km = no distance.
2. 100 km is twice as far as 50 km.
[Link]/Income
1. 0 salary = no money.
2. ₹40,000 is twice ₹20,000.

✅ Key Idea: Ratio = zero is real, and you can compare as multiples.
Simple Analogy

•Interval: Like a thermometer 🥛 → you can measure differences,

but zero doesn’t mean “nothing.”

•Ratio: Like a measuring scale 📏 → zero really means nothing, and

“twice as much” makes sense.
Question Type of Variable
Types of vehicle owned
Two wheeler, four wheeler
Product satisfaction
Unsatisfied, neutral, fairly satisfied, satisfied
To how many magazines do you currently
subscribed
Zero, One, Two, Three, Four
How tall are you (in inches)
Weight (in Kilograms)
Temperature (in degrees Celsius or degrees
Fahrenheit)
Population and Sample

 - Population: Entire group of individuals/data

points (e.g., all students in a university)
 - Sample: Subset taken from the population (e.g.,
200 students surveyed)
 Use Case: Election surveys where a sample
predicts population voting behavior.
Types of Sampling
Techniques
 1. Simple Random Sampling – Every
individual has equal chance
Example: Randomly picking lottery
tickets
 2. Stratified Sampling – Population divided
into groups, samples taken from each

Example: Sampling students by year (1st

year, 2nd year, etc.)
3. Cluster Sampling – Population divided
into clusters, some clusters chosen

Example: Selecting random schools in a

city, surveying all students in those schools

4. Systematic Sampling – Selecting every nth

individual

Example: Every 10th customer entering a

mall is surveyed
Desriptive
Vs
Inferential statistics
Definition:
Descriptive statistics means summarizing and
describing the data you already have.

It does not make predictions or

generalizations—just tells you what the data
shows.
Key Tools:
•Measures of Central Tendency (Mean,
Median, Mode)
•Measures of Dispersion (Range,
Variance, Standard Deviation)
•Tables, Charts, Graphs
Examples:
[Link] a class of 50 students, average score in maths =
72.
1. That’s descriptive—it describes the class.

[Link] a company, average salary of employees =

₹40,000.

1. Only about that company, no generalization.

[Link] showing number of people visiting a mall per
day.
1. Just describes what happened.

✅ In short: Descriptive = “Here’s what the data says.”

 Inferential
Definition:
Inferential statistics means making
conclusions or predictions about a
population using data from a sample.
It uses probability and hypothesis
testing.
Key Tools:
•Hypothesis Testing
•Confidence Intervals
•Regression, Correlation
•Chi-square, t-test, ANOVA
Examples:
1.A survey of 200 voters shows 60% support a candidate.
1. Inferential: Predicting that about 60% of the whole
population supports the candidate.

[Link] whether a new medicine is more effective than the old

one, based on a trial of 500 patients.
1. Generalizing results to all patients.

3.A company takes a sample of 100 transactions to check if fraud

occurs.
1. Inferential: Concluding the fraud rate in all transactions.

✅ In short: Inferential = “Using a sample to guess about the bigger

picture.”
Simple Analogy
•Descriptive Statistics = “Reading the report card” →
Tells you marks you got.

•Inferential Statistics = “Predicting future exams” →

Based on your marks, guess how you might do in the
final exam.
 Why Knowing Standard Deviation Matters

Reliability of the Average (Mean)

•Mean alone can be misleading.
•SD tells us if the mean is a trustworthy summary or not.

Example:
•Class A: Marks = [68, 69, 70, 71, 72] → Mean = 70, SD
=1.14(small)→ Mean represents veryone well.
•Class B: Marks = [20, 40, 70, 100] → Mean = 57.5, SD= 30(
large) → Mean does not represent most students.

SD helps us judge how “representative” the mean is.

Comparing Consistency
SD shows consistency (how stable data is).
Example:
•Factory A: Average machine output = 100 units, SD = 2
→ Very consistent.
•Factory B: Average output = 100 units, SD = 20 →
Highly inconsistent.

Even though averages are the same, Factory A is

much more reliable.
Risk and Uncertainty (Finance Example)
In investing:
•Stock X: Average return = 10%, SD = 2 → Stable,
predictable.
•Stock Y: Average return = 10%, SD = 15 → Risky,
big ups & downs.

Investors prefer smaller SD if they want stability.

Quality Control
In manufacturing:
•If SD of product size is small → almost all
products are close to the target size (good
quality).
•If SD is large → many defective or
unusable products.
Simple Analogy
Think of a cricket player’s runs in 5 matches:
•Player A: [48, 50, 52, 49, 51] → Average = 50,
SD small → Consistent performer.
•Player B: [0, 100, 5, 150, 0] → Average = 51, SD
large → Unpredictable performer.

Both have similar averages, but one is reliable,

the other is a hit-or-miss.
Why is CV Useful?
•Standard Deviation (SD) alone can’t always be
compared across datasets.
•CV lets you compare variability between datasets
with different scales or units.
•It answers: “Which dataset is more consistent
relative to its average?”
Examples
1️⃣ Factory Output
•Factory A: Mean = 100 units, SD = 5 → CV = 5%
•Factory B: Mean = 20 units, SD = 5 → CV = 25%
👉 Even though both have the same SD (5), Factory B is less consistent
because variation is huge compared to its small average.
Salaries in Companies
•Company A: Avg salary = ₹50,000, SD = ₹5,000 → CV =
10%
•Company B: Avg salary = ₹10,000, SD = ₹3,000 → CV =
30%
Company A is more stable in salary distribution,
Company B has more inequality.
Sports Performance
•Player A: Average runs = 50, SD = 10 → CV = 20%
•Player B: Average runs = 30, SD = 15 → CV = 50%
Player A is more consistent, Player B is more
unpredictable.
Types of Skewness
Zero Skewness (Symmetrical Distribution)
•Data is evenly spread on both sides of the mean.
•Mean ≈ Median ≈ Mode.
🎓 Examples:
•Heights of adults in a population (most are around
average).
•Marks in a well-designed exam (most students near
average, fewer very high/very low).
✅ Real Sense: Balanced distribution — the average is
a good summary.
Positive Skew (Right-Skewed)
•Tail is longer on the right (higher values).
•A few very large values “pull” the mean to the right.
•Mean > Median > Mode.

Examples:
•Income distribution: Most people earn moderate salaries, but a few
billionaires pull the mean up.
•House prices: Majority affordable, but a few luxury mansions push the
average higher.

Real Sense: Average looks higher than most people's actual values.
Negative Skew (Left-Skewed)
•Tail is longer on the left (lower values).
•A few very small values “pull” the mean down.
•Mean < Median < Mode.
Examples:
•Age at retirement: Most retire around 60–65, but a few retire very early
(30s/40s), pulling the mean lower.
•Exam with very easy paper: Most students score very high, but a few
score very low, dragging the average down.
Real Sense: Average looks lower than most people's actual values.
What is Kurtosis?

Kurtosis tells us about the "shape of the peak" of a distribution —

specifically, whether the data has heavy tails or light tails
compared to a normal distribution.
In plain words:
•Does the data have a lot of extreme values (outliers)?
•Or is it more flat and concentrated around the average?
Mesokurtic (Normal Kurtosis, ~3)
•Shape is “just right” (like a normal bell curve).
•Neither too peaked nor too flat.
•Few outliers.
Example:
•Heights of students in a class → most around
average, few extremes.
Real Sense: “Balanced” distribution.
Leptokurtic (High Kurtosis, >3)
•Sharper peak around the mean.
•Heavier tails → more extreme values/outliers.
•Data is tightly clustered in the middle, but extreme values happen more
often than normal.

Examples:
•Stock market returns: Most days small changes, but sometimes huge
jumps/drop.
•Exam marks where most score near 80–90, but a few score very low/high.

Real Sense: Data looks “peaked” in the middle, but surprises (outliers) are
more common.
Platykurtic (Low Kurtosis, <3)
•Flatter peak around the mean.
•Thinner tails → fewer outliers.
•Data is more evenly spread.

Examples:
•Uniform distribution (like rolling a fair dice many times).
•Exam where marks are spread evenly from 40–90.

Real Sense: Data is flat, no strong concentration in the

middle, and extreme values are rare.
What is Covariance?
Covariance tells us how two variables move together:
•Do they increase/decrease together?
•Or when one increases, does the other decrease?
•Or do they move independently (no relation)?
It’s like measuring the direction of the relationship
between two variables.
Interpretation of Covariance
[Link] Covariance
[Link] one variable increases, the other also increases.
Example:
[Link] studied and exam marks
[Link] hours studied → higher marks
Negative Covariance
•When one variable increases, the other
decreases.
Example:
•Speed of a car and travel time
•Higher speed → less time taken.
Zero Covariance
•No consistent relationship.

• Example:
•Shoe size and exam marks
•They don’t affect each other.