UNIT I: Descrip ve Sta s cs, Probability, and Discrete Probability Distribu ons 5.

Qualita ve Data
 Non-numeric data: categories or labels.
A. Descrip ve Sta s cs o Examples: Colors, Brands, Gender.
1. Popula ons and Samples  Displayed using bar charts, pie charts, etc.
 Popula on: The en re group of individuals or items of interest.
o Example: All students in a university. 6. Measures of Loca on (Central Tendency)
 Sample: A subset of the popula on selected for analysis.  Mean (μ or x̄): Average value.
o Used when studying the en re popula on is imprac cal.  Median: Middle value when ordered.
2. Stem-and-Leaf Displays  Mode: Most frequent value.
 Visual tool for quan ta ve data.
 Each value is split into: 7. Measures of Variability (Dispersion)
o Stem: Leading digit(s)  Range: Max - Min
o Leaf: Final digit  Variance (σ² or s²):
 Helps retain original data while showing distribu on. o Popula on:

Example: o Sample:
Data: 23, 24, 25, 26, 33, 34
Stem | Leaf  Standard Devia on (σ or s): Square root of variance.
2 |3456  Interquar le Range (IQR): Q3 - Q1 (middle 50%)
3 |34
8. Boxplots (Box-and-Whisker Plot)
3. Dotplots  Displays five-number summary: Min, Q1, Median (Q2), Q3, Max
 Each data value is shown as a dot above a number line.  Useful for detec ng skewness and outliers.
 Good for small datasets to observe clustering or gaps.
4. Histograms B. Probability
 Bar graph for quan ta ve data. 1. Sample Spaces and Events
 Divides data into intervals (bins) and plots frequency.  Sample Space (S): Set of all possible outcomes of an experiment.
 Useful to observe shape (normal, skewed), spread, and outliers. o Example: Tossing 2 coins → S = {HH, HT, TH, TT}
 Event: Any subset of the sample space.
2. Probability Axioms (Kolmogorov Axioms) Bayes' Theorem
For a sample space S, and event A ⊆ S : Bayes' Theorem relates condi onal probabili es and allows upda ng of probabili es based
on new evidence.
1. Non-nega vity:
Formula:
P(A)≥0
If B1,B2,…B_n are mutually exclusive and exhaus ve events and A is an event with P(A) > 0,
2. Normaliza on:
then:
P(S) = 1
3. Addi vity: If A ∩ B = ∅ (mutually exclusive),
P(A∪B)=P(A)+P(B)
Special Case:
3. Basic Proper es
 P(∅)=0
Use Case:
 P(A^c)=1−P(A)
Bayes' theorem is widely used in medical tes ng, spam filtering, and decision-making under
 If A ⊆ B , then P ( A ) ≤ P ( B ) uncertainty.

2. Independent Events
4. Condi onal Probability Two events A and B are independent if the occurrence of one does not affect the occurrence
 Probability of event A given event B has occurred: of the other.

P ( A ∣ B ) = P ( A ∩ B ) / P ( B ) If, P(B)>0 Mathema cal Condi on:

 Independent Events: P ( A ∩ B )= P ( A ) ⋅ P ( B )
A and B are independent if: For mul ple events A1 , A2 ,... An , mutual independence requires:
P ( A ∩ B ) = P ( A )⋅ P ( B )  P ( Ai ∩ Aj ) = P (Ai) P( Aj ) for all i≠j
 Extends to intersec ons of all subsets.
C. Discrete Probability Distribu ons (Outline Only for Con nuity)
While detailed distribu ons like binomial and Poisson come next in your syllabus, they 3. Discrete Random Variables and Probability Distribu ons
usually include:
Random Variable (RV):
 Binomial Distribu on
 A func on that assigns a real number to each outcome in the sample space.
 Poisson Distribu on
 Discrete RV: Takes on a countable number of values.
 Geometric Distribu on
Probability Distribu on Func on (PDF):
 Hypergeometric Distribu on
 For a discrete RV X, the probability mass func on (PMF) P (X = xi ) = pi where:
These models are used to describe the probabili es of outcomes in discrete scenarios.
∑ pi = 1 , 0 ≤ pi ≤ 1
Cumula ve Distribu on Func on (CDF):

F (x) = P (X ≤ x )
4. Expected Value (Mean) of a Discrete RV - d) Nega ve Binomial Distribu on
 Scenario: Number of trials to achieve r successes.
Other Key Measures:  PMF:
 Variance:  Mean:
 Variance:
 Standard Devia on:
e) Poisson Distribu on
5. Discrete Probability Distribu ons  Scenario: Number of occurrences in a fixed interval of me/space.
a) Binomial Distribu on  PMF:
 Scenario: Fixed number of independent trials, each with success probability p.  Mean = λ
 PMF:  Variance = λ
 Mean:
 Variance: 6. Poisson Distribu on as a Limit of Binomial
 When n → ∞ , p → 0 , and np = λ is constant:
b) Geometric Distribu on Binomial ( n , p ) ⟶ Poisson ( λ )
 Scenario: Number of trials un l the first success.  Jus fica on: Used when events are rare but trials are numerous (e.g., system
failures, radioac ve decay).
 PMF:

 Mean: UNIT II: Con nuous Probability Distribu ons

 Variance:

1. Con nuous Random Variables

c) Hypergeometric Distribu on A con nuous random variable (CRV) takes an infinite number of possible values in a given
 Scenario: Sampling without replacement from a finite popula on. interval.

 PMF:  Unlike discrete RVs, probability of any specific value is zero:

o N: Popula on size  Instead, we consider probability over intervals:

o K: Number of successes in the popula on

o n: Number of draws 2. Probability Density Func on (PDF)

o k: Number of observed successes A Probability Density Func on (PDF) is used to describe the probability distribu on of a
 Mean: con nuous random variable.
Proper es of a PDF: Cumula ve Distribu on Func on (CDF)

1. F ( x ) ≥ 0 for all x The Cumula ve Distribu on Func on (CDF) of a con nuous random variable XX is defined
as:
2. Total area under the curve = 1 :
∫−∞∞ f (x) dx = 1
Proper es of CDF:
3. Probability that X lies in an interval [a, b]:
 F(x)F(x) is non-decreasing
P ( a ≤ X ≤ b ) = ∫ f (x ) dx (over a,b)
 Lim x→−∞ F(x) = 0

 Lim x→∞ F(x )= 1

3. Cumula ve Distribu on Func on (CDF)  The PDF is the deriva ve of the CDF:
Defined as: F ( x ) = dF ( x ) / dx
F ( x ) = P ( X ≤ x ) = ∫ f ( t ) dt (over -infinity , x)  For interval probabili es:

 F(x) is non-decreasing. P( a < X ≤ b ) = F(b) - F(a)

 Lim x→−∞ F ( x ) = 0 , and lim x→∞ F ( x ) = 1

2. Expected Value (Mean) for Con nuous RV

4. Expected Value and Variance for Con nuous RVs Given a con nuous random variable X with PDF f(x):

 Expected Value (Mean):

3. Variance and Standard Devia on

 Variance:
 Variance:

 Standard Devia on:

5. Uniform Distribu on (Con nuous)

The Uniform distribu on is the simplest con nuous distribu on where all outcomes in an interval are equally likely.

PDF of Uniform Distribu on over [a, b]:

Mean and Variance: 4. Normal Distribu on

 Mean: The normal distribu on is the most important con nuous distribu on in sta s cs.
E(X)=a+b/2 Probability Density Func on (PDF):
 Variance:

Var (X) = (b−a)^2/12

 μ: Mean (center of the distribu on)
Graph:
 σ: Standard devia on (spread)
 Rectangular shape (constant height).
 Bell-shaped, symmetric about μ
 Total area under the curve is 1.

Example Applica on:

 If a bus arrives any me between 9:00 and 9:30 AM, the arrival me can be modeled using a uniform distribu on
over [0, 30].
 Standard Normal Distribu on (Z-distribu on): 6. Lognormal Distribu on

 Mean = 0, Standard devia on = 1 A random variable XX is lognormally distributed

PDF:

Use Z-tables to find probabili es.

Proper es:  Skewed to the right

 Symmetrical about the mean.  Defined only for x > 0

 Mean = Median = Mode. Proper es:

 The 68-95-99.7 Rule:  Mean:

o 68% of data within 1σ

o 95% within 2σ  Variance:

o 99.7% within 3σ

Applica ons:

5. Exponen al Distribu on  Modeling stock prices, income distribu ons, reac on mes, environmental data (e.g.,
pollu on levels)
Used to model me between events in a Poisson process.

PDF:

Range of
Distribu on Shape Mean Variance
Where λ > 0 is the rate parameter. X

CDF: Uniform a≤x≤ba Flat a+b/2 (b−a)^2 / 12

Normal −∞ <x <∞ Bell-shaped Μ σ^2

Proper es: Skewed
Exponen al X≥0 1/ λ 1 / λ^2}
 Mean: E (X) = 1 / λ right

 Variance: Var (X) = 1 / λ^2 Skewed Complicated

Lognormal x>0
right form
 Memoryless Property:

P(X>s+t∣X>s)=P(X > t)

– Past doesn't affect future probabili es.

Applica ons:

 Lifespan of components

 Time un l next phone call, machine failure, etc.

 4. Sca erplot of Bivariate Data

UNIT III: Central Limit Theorem and Regression Analysis  A sca erplot is a graph of paired data (x_i, y_i).

 Helps visualize:

1. Sampling Distribu on of the Sample Mean (X bar) o Direc on (posi ve/nega ve)

 When we repeatedly draw samples of size nn from a popula on and compute the sample o Form (linear/nonlinear)
mean X , the distribu on of these sample means is called the sampling distribu on of the
o Strength of rela onship
sample mean.
Interpreta on:
Key Points:
 Posi ve linear: Points rise from le to right.
 Mean of sampling distribu on:
 Nega ve linear: Points fall from le to right.
μX¯ = μ
 No correla on: Random pa ern.
 Standard devia on (Standard Error):

where σ is popula on standard devia on and n is sample size.

5. Regression Line using Least Squares Method
 As n increases, the sample mean becomes a more accurate es mate of the popula on mean.
Linear regression fits a straight line to the data using the least squares principle —
minimizing the sum of squared ver cal distances (errors) between actual and predicted values.
2. Central Limit Theorem (CLT)
Equa on of Regression Line:
The Central Limit Theorem is a founda onal theorem in sta s cs.
{y} = a + bx
Statement:
Where:
For a random sample of size nn from any popula on with mean μ\mu and standard devia on
 b: Slope =
σ , the sampling distribu on of the sample mean X¯ tends to be approximately normal as n becomes
large (typically n≥30), regardless of the shape of the popula on distribu on.  a: Intercept =

Implica ons: Interpreta on:

 Allows the use of normal probability models even when the original popula on is not  b: Change in y for each unit change in x
normal.
 a: Predicted value of y when x = 0
 Basis for confidence intervals and hypothesis tes ng.

3. Applica ons of CLT

6. Es ma on Using Regression Line
 Es ma ng popula on mean from sample mean using normal approxima on.
Once we have the regression equa on y^ = a + bx, we can:
 Construc ng confidence intervals for mean:
 Predict the value of y for any given x
 Hypothesis tes ng on means using z-tests and t-tests.
 Es mate the effect of changes in x on y
 Useful in quality control, polling, and risk modeling.
 7. Sample Correla on Coefficient (r)

Measures the strength and direc on of a linear rela onship between x and y.

Formula:

Proper es:

 −1 ≤ r ≤ 1

 r > 0 : Posi ve correla on

 r < 0 : Nega ve correla on

 r = 0 : No linear correla on

 |r| = 1: Perfect linear rela onship

Rela on to Slope (b):

If r > 0 , slope b > 0 ; if r < 0, slope b < 0