CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA.

Pranav Popat

Probability

→ First use of Probability was made 300 years back in Europe by a group of
Know about mathematicians to enhance their chances of winning in gambling
Probability → It is a full-fledged subject and become an integral part of statistics
→ Theories of Testing Hypothesis and Estimation are based on probability
Subjective Dependent on personal judgment, useful in decision making.
Probability It is out scope of our syllabus
Types
Objective This is based on Mathematical Rules and not judgment
Probability based. We will study this section in our chapter.
Experiment A performance that produces certain results
Random Random An experiment is defined to be random if the results of the
Experiment Experiment experiment depend on chance only.
Examples Tossing a coin, throwing a dice, drawing cards from a pack
Events The results or outcomes of a random experiment are known as events
Based on Combination of Events
Simple or Elementary If the event cannot be decomposed into
further events
Composite or Compound An event that can be decomposed into two or
more simple events

Based on nature of occurrence (applicable for set of events)

Mutually Exclusive or A set of events A1, A2 … is said to be mutually
Types of Events Incompatible Events exclusive if they cannot occur simultaneously.
Occurrence of one implies non occurrence of
other.
Exhaustive Events A set of events A1, A2 … is said to be
exhaustive if one of these must necessarily
occur on a random experiment
Equally Likely or Equi- If it is evident that from the set of events,
Probable Events or none of the events is expected to occur more
Mutually Symmetric frequently than others.
Also called Prior Definition of Probability, this formula is Event (Result) Based.
Classical
It is given by Bernoulli and Laplace.
Definition of no. of events favorable to A
Probability 𝑃(𝐴) =
total number of events

1 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Demerits or → Applicable only when events are finite and are equally
Limitations likely
→ Limited application of this definition like in tossing coin,
throwing dice, cards etc.
Other Notes → 0 ≤ 𝑃(𝐴) ≤ 1, P(A) = 1 means Sure Event, P(A) = 0
More about means impossible event
Classical → Probability of non-occurrence of an event A is denoted
Probability by P(A’) or P(A) is called as complimentary event of A.
P(A’) = 1−P(A)
Odds in Favor no. of favorable events
of an Event no. of unfavorable events
Odds Against no. of unfavorable events
an Event no. of favorable events
If an experiment results in p outcomes and if it is repeated q times then
Special Formula
Total no. of outcomes = 𝑝
Suits
Terms used in 52
Cards Deck
(four) Spades - ♠ Hearts - Diamond - Clubs - ♣
Ranks
A (Ace), K (King), Q (Queen), J (Jack), 10, 9, 8, 7, 6, 5, 4, 3, 2
(13)
no.of times the event occured during experimental trials 𝑓𝐴
Relative Frequency Relative Frequency = total no.of trials
= n
Definition of 𝑓
Probability by this method is defined as 𝑃(𝐴) = lim → n𝐴
Probability
(Relative Frequency on infinite no. of trials is equal to probability)
Sample Space (denoted a non-empty set containing all the elementary
by S or Ω-omega) events of a random experiment as sample points
Event A Event which is under consideration for probability
Set Based calculations is defined as a non empty subset of Set
Probability S (Sample Space)
Probability Formula no. of sample points in A 𝑛(𝐴)
𝑃(𝐴) = =
no. of sample points in S 𝑛(𝑆)

This definition is also based on Sets Concepts. Here Probability is not a simple
ratio like above, but can be said as function P defined on S known as
Probability Measure.
Axiomatic Or P(A) is defined as the probability of A as per this function only if below
Modern conditions are satisfied:
Definition of Condition 1 𝑃(𝐴) ≥ 0, for every 𝐴 ⊆ 𝑆
Probability Condition 2 𝑃(𝑆) = 1
Condition 3 For any sequence of mutually exclusive events A1, A2, A3...
𝑃(𝐴 ∪ 𝐴 ∪ 𝐴 ∪ … ) = 𝑃(𝐴 ) + 𝑃(𝐴 ) + 𝑃(𝐴 ) + …

2 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Theorem 1 𝑃(𝐴 ∪ 𝐵) = 𝑃(𝐴 + 𝐵) = 𝑃(𝐴 𝑜𝑟 𝐵) = 𝑃(𝐴) + 𝑃(𝐵)

If A and B are mutually exclusive events
Theorem 2 For set of mutually exclusive events 𝐴 , 𝐴 , 𝐴 , …
Addition 𝑃(𝐴 ∪ 𝐴 ∪ 𝐴 ∪ … ) = 𝑃(𝐴 ) + 𝑃(𝐴 ) + 𝑃(𝐴 ) + …
Theorems Theorem 3 𝑃(𝐴 ∪ 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵)
For any two events A and B
Theorem 4 𝑃(𝐴 ∪ 𝐵 ∪ 𝐶) = 𝑃(𝐴) + 𝑃(𝐵) + 𝑃(𝐶) − 𝑃(𝐴 ∩ 𝐵)
− 𝑃(𝐵 ∩ 𝐶) − 𝑃(𝐴 ∩ 𝐶) + 𝑃(𝐴 ∩ 𝐵 ∩ 𝐶)
Expected
No. of sample points n(S) × P(A)
Frequency
Dependent If occurrence of one event is influenced by occurrence of
Events another event, then two events are dependent.
Independent Two events are said to be independent if occurrence of
Events one event do not influence the occurrence of other.
Probability in case Conditional Probability of B/A: means probability of event B
of Dependent given that event A has already been occurred
Events A and B 𝐵 𝑃(𝐵 ∩ 𝐴)
𝑃 = , provided 𝑃(𝐴) > 0
𝐴 𝑃(𝐴)
Similarly, Conditional Probability of A/B:
𝐴 𝑃(𝐵 ∩ 𝐴)
𝑃 = , provided 𝑃(𝐵) > 0
𝐵 𝑃(𝐵)
Conditional Compound Theorem: P(A∩B) = P(B) × P(A/B) or
Probability or P(A∩B) = P(A) × P(B/A)
Compound Probability in Since there is no dependency,
Theorem case of Conditional Probability = Normal Probability
Independent
Events i.e. P(B/A) = P(B) and P(A/B) = P(A)
Here, P(A∩B)= P(A) × P(B)

And for three events, A, B, C

P(A∩B∩C) = P(A) × P(B) × P(C)

Also, if A and B are independent, then below are also

independent :
A and B’, A’ and B, A’ and B’

Random Variable It is a function defined on Sample Space of a random

experiment that can take any value (Real Number)
Discrete Random RV that can take only discrete values. RV on a discrete
Variable sample space
Continuous RV that can take any values within an interval. [infinite
Random Variable:
Random Variable no. of sample points in a sample space]
Probability
Probability If is defined as the statement/ table that shows no. of
Distribution
Distribution different value taken by Random Variable and their
corresponding probabilities
Conditions of If X (a random variable) takes n finite values like
Probability Dist. 𝑋 , 𝑋 , 𝑋 , … , 𝑋 and probabilities are
𝑃 , 𝑃 , 𝑃 , … , 𝑃 then, 𝑃 ≥ 0 for every 𝑖 𝑎𝑛𝑑 ∑ 𝑃 = 1

3 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Expected It is defined as the sum of products of different values taken

Value by Random Variable and corresponding probabilities.
𝐸(𝑥) = ∑ 𝑝 𝑥 (this formula is similar to AM of frequency distribution)
Mean of Since this is mean, we can say that Expected value is equal to
Probability arithmetic mean of probability distribution. Here mean is
Distribution denoted by μ, hence 𝜇 = 𝐸 (𝑥 ) = ∑ 𝑝 𝑥
Expected Value Variance of
Probability 𝑉(𝑥) = 𝜎 = 𝐸(𝑥 − 𝜇) = 𝐸(𝑥) − 𝜇
Distribution
Properties → E.V. of a constant is constant
of E.V. → 𝐸(𝑥 + 𝑦) = 𝐸(𝑥) + 𝐸(𝑦)
→ 𝐸(𝑘. 𝑥) = 𝐸(𝑥). 𝑘
→ 𝐸(𝑥. 𝑦) = 𝐸(𝑥). 𝐸(𝑦)

4 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Theoretical Distribution

Bernoulli’s → Each trial is associated with two mutually exclusive

Trial and exhaustive outcomes [one is success and other one
is failure]
→ Trials are independent
→ Probability of success (p) and failure (q=1-p) will
remain unchanged throughout the process
→ No. of trials is a positive integer
Binomial It is a discrete random variable X that follows binomial
Variable distribution and is denoted by 𝑿~𝑩(𝒏, 𝒑)
Binomial Probability 𝒇(𝒙) = 𝑷(𝑿 = 𝒙) = 𝒏𝑪𝒙 𝒑𝒙 𝒒𝒏 𝒙
Distribution Mass Function for 𝑥 = 0,1,2,3, … , 𝑛 and 𝑓(𝑥) = 0 if 𝑥 is otherwise
(bi-parametric discrete
probability
Mean 𝝁 = 𝒏𝒑
distribution) Variance 𝟐
𝝈 = 𝒏𝒑𝒒, also Variance is always less than mean,
maximum value of variance is 𝑛 4
Mode Calculate (𝑛 + 1)𝑝,
if the resulting value is 𝝁𝟎 = (𝑛 + 1)𝑝 𝑎𝑛𝑑
integer then Bi-modal [(𝑛 + 1)𝑝 − 1]
If the resulting value is non- 𝝁𝟎 = largest integer
integer then Uni-modal contained in (𝑛 + 1)𝑝
Additive If X and Y are two independent variables such that
Property 𝑿~𝑩(𝒏𝟏 , 𝒑) and 𝒀~𝑩(𝒏𝟐 , 𝒑), then (𝑿 + 𝒀)~𝑩(𝒏𝟏 + 𝒏𝟐 , 𝒑)

Simon Denis Poisson of France introduced this

History
distribution way back in the year 1837
It is a limiting form of Binomial Distribution, where n→∞,
Conditions
p→0. It is also a discrete distribution
Poisson It is a discrete random variable that follows Poisson
Variable Distribution denoted as 𝑿~𝑷(𝒎)
𝒎
(𝒆 . 𝒎𝒙 )
Probability 𝒇(𝒙) = 𝑷(𝑿 = 𝒙) =
Poisson Mass Function 𝒙!
Distribution for 𝑥 = 0, 1, 2, … ∞
(uni-parametric Mean
discrete probability
𝝁=𝒎
distribution) Variance 𝝈𝟐 = 𝒎
Mode Calculate 𝑚,
if the resulting value is 𝝁𝟎 = 𝒎 𝑎𝑛𝑑
integer then Bi-modal [𝒎 − 𝟏]
If the resulting value is non- 𝝁𝟎 = largest integer
integer then Uni-modal contained in 𝒎
Additive If X and Y are two independent variables such that
Property 𝑿~𝑷(𝒎𝟏 ) and 𝒀~𝑷(𝒎𝟐 ), then (𝑿 + 𝒀)~𝑷(𝒎𝟏 + 𝒎𝟐 )

5 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Basics Various Mathematical experiments have proved that most

Normal of the continuous random variables will follow normal
Distribution distribution. It is universally accepted distribution.
(bi-parametric Probability 1 ( ) ×
continuous probability Density 𝑓(𝑥) = ×𝑒
distribution) 𝜎√2𝜋
Function
It is defined for -∞ < 𝑥 < ∞

Mean = Median = Mode μ

Standard Deviation σ
Mean Deviation 𝜎 × 2/𝜋 = 0.8 𝜎
Quartile Deviation Q1= 𝜇 − 0.675𝜎 and Q3 = 𝜇 + 0.675𝜎
Shape of Normal Curve Bell Shaped
Normal Normal Variable 𝑋~𝑁(𝜇, 𝜎 )
Distribution Only applicable when two different random
Properties variables are independent. Assume we have two
Additive Property variables X and Y such that 𝑿~𝑵(𝝁𝟏 , 𝝈𝟏 𝟐 ),
𝒀~𝑵(𝝁𝟐 , 𝝈𝟐 𝟐 ) then X+𝒀~𝑵(𝝁𝟏 + 𝝁𝟐 , 𝝈𝟏 𝟐 + 𝝈𝟐 𝟐 )
Normal Curve is 𝑥=𝜇
symmetrical at
Points of Inflexion 𝜇−𝜎&𝜇+ 𝜎
Ratio between QD:MD:SD 10:12:15
Conditions Parameter Value
Mean μ 0
Standard Deviation σ 1
Standard Normal Variate The variable used in this distribution is called
as Standard Normal Variate and is denoted by
𝑍 [Striked Z]
Area from X=-3σ to X=3σ 99.73%
Z Table This table gives us the probability of values
from X=μ=0 to X=any value up to 3
Z Score 𝑥−𝜇
Standard Normal 𝑍=
𝜎
Distribution Cumulative Distribution
𝜙(𝑥) = 𝑃(𝑋 ≤ 𝑥
Function
Probability Function 𝑓(𝑧) = 𝑒 ( ) ×
for -∞<z< ∞
√
Mean, Median, Mode μ=0
SD, Variance σ=1, σ2=1
Points of Inflexion -1, 1
Mean Deviation 0.8
Quartile Deviation 0.675
Probability Function 𝑓(𝑧) = 𝑒 ( ) ×
for -∞<z< ∞
√

6 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – ST
TATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Area under
Normal Curve
From To Area/Probability
μ +σ 34.135%
+σ +2σ 13.59%
+2σ +3σ 2.14%
3σ ∞ 0.135%

From To Area/Probability
-σ +σ 68.3%
-2σ +2σ 95.5%
-3σ +3σ 99.7%

7 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat,
pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

A. INTRODUCTION TO STATS

Definition
Singular Sense:
 Scientific method that is used for collecting, analyzing and presenting data
 Used to draw statistical inferences
 Inferences means conclusion reached on the basis of evidence and reasoning
Example:
After applying statistical methods we have arrived at a conclusion that in last 5 years crime rate is
reduced.
Plural Sense:
 Data qualitative or quantitative collected to do statistical analysis
Example: Based on Cricket Match statistic of this stadium, chasing team wins mostly
History of Stats
 Word Origin
 Latin word – Status
 Italian word – Statista
 German word – statistic
 French word – statistique
 Publication:
 Koutilya’s book Arthashastra
 Stat records on Agriculture found in Ain-i-Akbari (author Abu Fezal)
 Census: First ever census done in Egypt (300 years BC to 2000 BC)
Application of Stats
There are various but we will confine to below:
1. Economics: Time Series analysis, index, demand analysis, econometrics, regression analysis
2. Business Management: business decisions rely upon QT
3. Commerce/ Industry: Sales, Purchase, RM, Salary Wages etc. data are analyze for business
decisions and policy making
Limitation of Stats:
1. Relevant for aggregate data and not individual data
2. Quantitative data can only be used, however for qualitative – it needs to be converted into
quantitative
3. Projections are based on conditions/ assumptions and any change in that will change the
projection
4. Sampling based conclusions are used, improper sampling leads to improper results

8 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

B. COLLECTION OF DATA

Data and Variable

 Variable = measurable quantity
 Discrete variable: when a variable assumes a finite or count ably infinite isolated values.
Example: no. of petals in a flower, no. of road accident in locality
 Continuous variable: when a variable assumes any value from the given interval (can
also be in decimals, fractions). Example: height, weight, sale, profit
 Attribute: qualitative characteristics. Example: Gender of a baby, nationality of a person
 Data = quantitative information shown as number. These are of two types:
 Primary : first time collected by agency/ investigator
 Secondary: collected data used by different person/ agency

How to collect Primary Data?

1. Interview Method:
a. Personal Interview: directly from respondents. Example: Natural Calamity, Door to
Door Survey
b. Indirect Interview: when reaching to person difficult, contact associated persons.
Example: Rail accident
c. Telephone Interview: over phone, quick and non-responsive

Type of Interview/
Personal Indirect Telephone
Parameters
Accuracy High Low Low
Coverage Low Low High
Non Response Low Low High

2. Mailed Questionnaire Method:

a. Mailed means by Post or Email
b. Well drafted + properly sequenced + with guidelines
c. Non Response is Maximum

3. Observation Method:
a. Data collected by direct observation or using instrument
b. Example: Height check, Weight check,
c. Although more accurate but it is time consuming, low coverage and laborious

4. Questionnaire filled and sent by Enumerators

a. Enumerator: Person who directly interact with respondent and fill the questionnaire
b. Generally used in Surveys
Sources of Secondary Data
1. International sources like World Health Organization (WHO), International Monetary Fund
(IMF), International Labor Organization (ILO), World Bank
2. Government Sources – In India – Central Statistics Office (CSO), National Sample Survey
Office- NSSO, Regulators – RBI, SEBI, RERA, IRDA
3. Private or Quasi-government sources like Indian Statistical Institute (ISI), Indian Council of
Agriculture, NCERT
4. Research Papers and other unpublished sources

9 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – ST
TATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Scrutiny of Data

1. Scrutiny – checking accuracy and consistency of data

2. Finding of errors by enumerators while filling or receiving questionnaire
3. Internal consistency check: when two or more series of related data are given check each
other
4. Consider enumerators’ bias while using data

C. PRESENTATION OF DATA

Classification and organization of Data

Data:
 means process of arranging data based on some logic
 there are four types of classification of data
a. Chronological/ Temporal/ Time Series Data (ex. Profit YoYi.e year on year)
b. Geographical or Spatial Series Data (ex. Weather in North India and South India)
c. Qualitative or Ordinal Data (ex. Rating Top 20 songs by Radio Mirchi)
d. Quantitative or Cardinal Data (no. of left handed batsmen in cricket teams playing
CWC19)
Mode of Presentation
1. Textual: where text is used in the form of para or sentence. Example: Height of A,B and C is
160cm, 165cm,
cm, 175cm respectively
2. Tabular/ Tabulation:
 Data shown in the form of table
 Some important terms about Table (we will understand by example - next page figure)
 It is preferred over textual form because
 Useful in easy comparison
 Complicated data can be presented
 Table is must to create a diagram
 No analysis possible without diagram

10 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat,
pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

3. Diagrammatic representation of data

 Can be helpful for layman (without having much knowledge of numbers)
 Hidden trend can be traced
 Table is more accurate than diagrams
 Types of Diagram below:

Line Diagram/ Histogram:

 plotting points in graph and join them to make a line
 used generally for time series (variable y is plotted against time t)
 for wide fluctuation, log chart or ratio chart is used (log y is plotted against t)
 for two or more series of same unit – multiple line chart is used
 for two or more series of distinct unit – multiple axes chart is used
 Refer Material for Diagram
Bar Diagram
 Bar means rectangle of same width and of varying length drawn horizontally or
vertically
 For comparable series – multiple or grouped bar diagrams can be used
 For data divided into multiple components – subdivided or component bar diagrams
 For relative comparison to whole, percentage bar diagrams or divided bar diagrams
Pie Chart
 Used for circular presentation of relative data (% of whole)
 Summation of values of all components/segments are equated to 360 Degree (total
angle of circle)
°
 Segment angle =

11 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

D. FREQUENCY DISTRIBUTION

What is Frequency Distribution?

Frequency means number of times a particular observation is repeated. This applies to both variable
and attribute. It is shown in tabular form with class interval or the observation in one column and its
frequency in the other.
These are of two types
 Ungrouped/ Simple Frequency Distribution
 Grouped Frequency Distribution

Important Terms

1. Mutually exclusive classification or Overlapping Classification: This is usually applicable for

continuous variable. An observation as UCL is excluded from the class interval and taken in the
class where it is LCL.
Example: in the below class interval where will the observation 20 fall?
Class Class where
20 will fall
10-20 No – excluded
20-30 Yes
30-40 No

2. Mutually inclusive classification or Non Overlapping Classification: This is usually

applicable to discrete variable. All observation including UCL and LCL will be taken in the same
class interval as there is no confusion.
Example:
Class Class where
20 will fall
10-19 No
20-29 Yes
30-39 No

3. Class Limit: for a class interval CL is the minimum and maximum value the class interval may
contain. Minimum = Lower Class Interval (LCL) and Maximum = Upper Class Interval (UCL)
Example:

Class Type LCL UCL Class Type LCL UCL

10-19 Mutually Inclusive 10 19 10-20 Mutually Exclusive 10 20
20-29 Mutually Inclusive 20 29 20-30 Mutually Exclusive 20 30
30-39 Mutually Inclusive 30 39 30-40 Mutually Exclusive 30 40

4. Class Boundary: These are actual class limits of a class interval

a. For Mutually Exclusive / Overlapping : Class Boundary = Class Limit
LCL = LCB, UCL = UCB
b. For Mutually Inclusive / Non Overlapping: Mid of the two class limits
LCB = LCL – D/2, UCB = UCL + D/2

12 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Example:

Class Type LCL UCL LCB UCB Class Type LCL UCL LCB UCB
Mutually Mutually
10-19 10 19 9.5 19.5 10-20 10 20 10 20
Inclusive Exclusive
Mutually Mutually
20-29 20 29 19.5 29.5 20-30 20 30 20 30
Inclusive Exclusive
Mutually Mutually
30-39 30 39 29.5 39.5 30-40 30 40 30 40
Inclusive Exclusive

5. Mid Point/ Mid Value of Class / Class Mark

𝐋𝐂𝐋 𝐔𝐂𝐋 𝐋𝐂𝐁 𝐔𝐂𝐁
or
𝟐 𝟐
6. Width / Size of Class Interval
UCB – LCB

7. Cumulative Frequency

Class Frequency Less than type CF More than type CF

10-20 5 5 18
20-30 2 7 13
30-40 8 15 11
40-50 3 18 3
Total 18

8. Frequency Density
𝐅𝐫𝐞𝐪𝐮𝐞𝐧𝐜𝐲 𝐨𝐟 𝐜𝐥𝐚𝐬𝐬
𝐂𝐥𝐚𝐬𝐬 𝐥𝐞𝐧𝐠𝐭𝐡 𝐨𝐟 𝐭𝐡𝐚𝐭 𝐜𝐥𝐚𝐬𝐬
9. Relative Frequency or % Frequency

𝐅𝐫𝐞𝐪𝐮𝐞𝐧𝐜𝐲 𝐨𝐟 𝐜𝐥𝐚𝐬𝐬
𝐓𝐨𝐭𝐚𝐥 𝐅𝐫𝐞𝐪𝐮𝐞𝐧𝐜𝐲 𝐨𝐟 𝐭𝐚𝐛𝐥𝐞
Class Frequency Relative Percent
Class Frequency
Length Density Frequency Frequency
10-20 5 10 0.5 5/18 27.7%
20-30 2 10 0.2 2/18 11.11%
30-40 8 10 0.8 8/18 44.44%
40-50 3 10 0.3 3/18 16.67%
Total 18

Graphical Presentation of Frequency Distribution

1. Histogram/ Area Diagram [refer study material page 14.20 for diagram]
a. It is a convenient way to represent FD
b. Comparison between frequency of two different classes possible
c. It is useful to calculate mode also
d. Steps to create
 Covert CL into CB and plot in x axis

13 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

 Form rectangles taking class interval as base (x axis)

 And frequency as length (y axis) | Use frequency density in case of uneven length

2. Frequency Polygon
a. Usually preferable for ungrouped frequency distribution
b. Can be used for grouped also but only if class lengths are even
c. Steps to create
 Plot (xi, fi) where xi = class value (in case of ungrouped), mid value (in case of
grouped) and fi = frequency
 Join all plotted points to make line segments which eventually will become a polygon
(a shape with multiple number of line segments)

3. Ogives/ Cumulative Frequency Graph

a. Create a table where cumulative frequency is mapped against each CB (Class Boundary) and
make a curve by plotting and joining points by line segments. (curve is called Ogive)
b. This graph can be made by both type of Cumulative Frequency and called as Less than Ogive
or More than Ogive
c. It can be used for calculating quartiles also
d. If we plot both ogives in same graph, perpendicular line drawn from their intersection
towards x axis is cutting axis at Median

4. Frequency Curve
a. It is a limiting form of Area Diagram (Histogram) or frequency polygon
b. It is obtained by drawing smooth and free hand curve though the mid points
c. These are of below four types:
 Bell Shaped
 U-Shaped
 J-Shaped
 Combination of Curves as Mixed Curve

14 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Central Tendency

Central Tendency is the tendency of a given set of observations to cluster around a

Meaning single central or middle value and the single value that represents the given set of
observations is described as a measure of central tendency or, location, or average.

Definition the sum of all the observations divided by the number of

observations
Formula for ⋯
discrete 𝑥̅ = or
distribution
Formula for
frequency
𝑥̅ =
distribution N = Σ𝑓, 𝑥 = mid-point in case of grouped frequency distribution
Arithmetic
( )
Mean Deviation 𝑥̅ = 𝐴 + × 𝐶, where d =
Method
A = assumed mean, C = class length
→ If all the observations are constant, AM is also constant
→ the algebraic sum of deviations of a set of observations
from their AM is zero
Properties
→ AM is affected both due to change of origin and scale
̅ ̅
→ Combined Mean: 𝑥̅ =

Definition the middle-most value when the observations are arranged

either in an ascending order or a descending order of magnitude
For Discrete Step 1: Arrange data in ascending (or descending) order
Distribution
Step 2: Use the formula term

Step 1: Prepare a less than type cumulative frequency

distribution with Class boundaries as base.
For Frequency
Median (one Distribution Step 2: Calculate N/2 and check between which class
of the boundaries it falls. Mark LCB as 𝑙 and 𝑙 and corresponding
partition (refer example cumulative FD as 𝑁 and 𝑁
values) 15.1.6 – Page
15.8 Study Step 3: Apply the below formula
Mat)
−𝑁
𝑀𝑒 = 𝑙 + × 𝐶𝑙𝑎𝑠𝑠 𝑙𝑒𝑛𝑔𝑡ℎ
𝑁 −𝑁

→ Median is affected by both change of origin and scale

→ For a set of observations, the sum of absolute deviations
Properties
is minimum, when the deviations are taken from the
median.

15 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Meaning values dividing a given set of observations into a number of

equal parts
Median Median is also a quartile that divides the set of observations into
two equal parts.
Number of equal parts Four (4)
Quartiles Number of Quartiles Three (3)
Denoted by 𝑄 ,𝑄 ,𝑄

Number of equal parts Ten (10)

Deciles Number of Deciles Nine (9)
Denoted by 𝐷 ,𝐷 ,𝐷 ,…,𝐷

Partition
Values Number of equal parts Hundred (100)
Percentiles Number of Percentiles Ninety Nine (99)
Denoted by 𝑃 ,𝑃 ,𝑃 ,…,𝑃

(𝑛 + 1)𝑃 𝑡𝑒𝑟𝑚,
𝑃 Quartile 1 2 3
ℎ𝑒𝑟𝑒 𝑝 = , , ,
4 4 4
How to
calculate (𝑛 + 1)𝑃 𝑡𝑒𝑟𝑚,
𝑃 Decile 1 2 3 9
Partition ℎ𝑒𝑟𝑒 𝑝 = , , ,…,
Values 10 10 10 10
(𝑛 + 1)𝑃 𝑡𝑒𝑟𝑚,
𝑃 Percentile 1 2 3 99
ℎ𝑒𝑟𝑒 𝑝 = , , ,…,
100 100 100 100

Definition Mode is the value that occurs the maximum number of times.
Type of Mode A distribution can be uni-modal, bi-modal or multi-modal
𝑓 −𝑓
𝑀𝑜𝑑𝑒 = 𝑙 + × 𝐶𝑙𝑎𝑠𝑠 𝑙𝑒𝑛𝑔𝑡ℎ
Mode 2𝑓 − 𝑓 − 𝑓
For Frequency
Distribution Here,
𝑓 = 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑚𝑜𝑑𝑎𝑙 𝑐𝑙𝑎𝑠𝑠,
𝑓 = 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑜𝑓 𝑝𝑟𝑒 − 𝑚𝑜𝑑𝑎𝑙 𝑐𝑙𝑎𝑠𝑠,
𝑓 = 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑜𝑠𝑡 𝑚𝑜𝑑𝑎𝑙 𝑐𝑙𝑎𝑠𝑠

Empirical For a moderately skewed distribution,

Relationship 𝑀𝑒𝑎𝑛 – 𝑀𝑜𝑑𝑒 = 3 × (𝑀𝑒𝑎𝑛 – 𝑀𝑒𝑑𝑖𝑎𝑛)

16 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

For a given set of n positive observations, the geometric mean is

Definition
defined as the 𝑛 root of the product of the observations
Formula 𝐺 = (𝑥 × 𝑥 × 𝑥 … × 𝑥 ) /
Geometric → log 𝐺 = Σ log 𝑥
Mean
→ If all observations are constant GM is also constant
Properties → 𝐺𝑀 𝑜𝑓 𝑥𝑦 = 𝐺𝑀 𝑜𝑓 𝑥 × 𝐺𝑀 𝑜𝑓 𝑦
→ 𝐺𝑀 𝑜𝑓 =

For a given set of non-zero observations, harmonic mean is

Definition defined as the reciprocal of the AM of the reciprocals of the
observation
𝑛
Formula 𝐻=
Harmonic Σ(1 𝑥 )
Mean
→ If all observations are constant HM is also constant
Properties → Combined HM:

In case of rates like speed, hours per HM is used

When to use
day, etc.
GM and HM
In case of % and ratios GM is used

Relationship General AM ≥ GM ≥ HM
between AM, When all the observations are same AM = GM = HM
GM and HM When all the observations are distinct AM > 𝐺𝑀 > HM

Best Measure – Overall AM

Best Measure for Open End Class Median
Ideal Measure Based on all observations AM, GM, HM
of Central Based on 50% values Median
Tendency Not affected by Sampling fluctuations Median
Rigidly defined, easy to comprehend AM, Median, GM, HM
No Mathematical Property Mode

17 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Dispersion

Dispersion for a given set of observations may be defined as the amount of

Definition deviation of the observations, usually, from an appropriate measure of central
tendency
Absolute These are with units and not useful for comparison of two
Measures of variables with different units. Example: Range, Mean
Dispersion Deviation, Standard Deviation, Quartile Deviation
Types of
These are unit free measures and useful for comparison of
Dispersion
Relative Measures two variables with different units. Example: Coefficient of
of Dispersion Range, Coefficient of Mean Deviation, Coefficient of
variation, Coefficient of Quartile Deviation

Definition Difference between the largest and smallest of observations.

Formula Range = L – S
Relative Coefficient of Range = × 100
Range Measure
→ No effect of change of origin but affected by change of
Properties scale in the magnitude (ignore sign).

Mean deviation is defined as the arithmetic mean of the

Definition absolute deviations of the observations from an appropriate
measure of central tendency
1
𝑀𝐷 = Σ|𝑥 − 𝐴|
Mean Formula 𝑛
Deviation Here A is mean or median as given in question
Relative Mean Deviation about A
Measure Coefficient of Mean Deviation =
A
× 100
→ No effect of change of origin but affected by change of
Properties
scale in the magnitude (ignore sign).

It is defined as the root mean square deviation when the

Definition
deviations are taken from the AM of the observations
Σ(𝑥 − 𝑥̅ ) Σ𝑥
Formula 𝑆𝐷 𝑜𝑟 𝜎 = 𝑜𝑟 − (𝑥̅ )
n 𝑛
Relative SD
Measure
Coefficient of Variation =
AM
× 100
| |
Standard For any two numbers, a and b SD =
Deviation Standard Result
(𝑛 − 1)
SD of first n natural numbers
12
→ If all the observations are constant, SD is Zero
→ No effect of change of origin but affected by change of
scale in the magnitude (ignore sign)
Properties of SD
2 2
𝑛1𝑠1 2+𝑛2𝑠22 +𝑛1𝑑1 +𝑛2𝑑2
→ Combined SD = 𝑛1 +𝑛2

18 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Definition It is defined as the semi-inter quartile range

𝑄 −𝑄
Formula 𝑄 =
Quartile 2
Deviation
Relative 𝑄3 −𝑄1
Measure
Coefficient of Quartile Deviation =
𝑄3 +𝑄1
× 100

Best Measure – Overall SD

Best Measure for Open End Class QD
Quickest to compute Range
Ideal Measure Not based on all observations Range
of Dispersion Difficult to comprehend and less Mean Deviation
Mathematical
Rigidly defined, easy to comprehend Mean Deviation, SD, QD
Not affected by Sampling fluctuations QD

19 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

CORRELATION

Bi-Variate When data are collected on two discrete variables simultaneously, they are known
Data as Bi-Variate data
Bi-Variate
Distribution of Bi-Variate data is called as Bivariate Distribution
Distribution
Meaning Frequency distribution involving two discrete variables.
Marginal If we make a separate distribution from bi-variate frequency
Distribution distribution where we take aggregate of only one variable at a
Bi-Variate
time. Total no. of marginal distributions = 2
Frequency
Conditional If we make a separate distribution from bi-variate frequency
Distribution
Distribution distribution where we take one variable related one class
interval of another variable. Total no. of conditional
distributions = m + n (m = no. of rows, n = no. of columns)

While studying two variables at the same time, if it is found that the change in one
variable leads to change in the other variable either directly or inversely, then the
two variables are known to be associated or correlated.
Correlation
Positive Correlation If two variables move in the same direction
Negative Correlation If two variables move in the opposite direction
No Correlation If no change due to each other

A measurement or formula that represents the nature/ direction and/or

magnitude of correlation.
Method Helps in obtaining
Scatter Diagram Only direction of correlation
Karl Pearson’s Product Direction as well as strength of correlation.
Measure of
moment correlation coefficient Best Method – Most accurate
Correlation
Spearman’s rank correlation Direction as well as strength of correlation.
co-efficient Useful for attributes.
Co-efficient of concurrent Direction as well as strength of correlation.
deviations Only preferred for direction and not
magnitude. Quickest method.

20 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – ST
TATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Scatter
Diagram

the ratio of covariance between the two variables to the

Defined as
product of the standard deviations of the two variables
𝐶𝑜𝑣(𝑥, 𝑦)
Main Formula 𝑟 =
𝜎 .𝜎

Formula for Σ(𝑥 − 𝑥̅ )(𝑦 − 𝑦) Σ𝑥𝑦

𝑥𝑦
Covariance 𝐶𝑜𝑣(𝑥, 𝑦) = 𝑜𝑟 − 𝑥̅ . 𝑦
𝑛 𝑛
Karl Pearson’s Formula for Standard Σ(𝑥 − 𝑥̅ ) Σ𝑥
Product moment Deviation 𝜎 𝑜𝑟 𝜎 𝜎 = 𝑜𝑟 − ((𝑥̅ )
correlation coefficient n 𝑛

→ It is a unit-free measurement
→ Value of r lies from -1 to +1 both inclusive
→ Change of origin or Scale
Properties Change of Origin No impact
Change of Scale No impact of value, but if change of
scale of both variables are of different
sign then sign r will also change

find the level of agreement

ement (or disagreement) between
Applied to two judges so far as assessing a qualitative
characteristic is concerned
2
6Σ𝑑
Main Formula 𝑟 = 1− , here d means difference in ranks
Spearman’s 𝑛 𝑛2 −1
Rank
Correlation coefficient Adjustment Value in Σ here t is a tie length and we need to do
case of Tie Rank
summation of all ties
Formula in case of Tie ( )
length
𝑟 = 1− ( )

21 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat,
pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

A very simple and casual method of finding correlation

Use when we are not serious about the magnitude of the
two variables
This method involves in attaching a positive sign for a x-value
(except the first) if this value is more than the previous value
and assigning a negative value if this value is less than the
Steps in this method previous value. Applies to both variable and then these signs
Co-efficient of
concurrent are compared. If signs match – pair is counted as concurrent
deviations deviation.
2𝑐 − 𝑚
𝑟 = ± ±
𝑚
Formula
Here, m = total no. of deviations (it is one less than total
no. of pairs under observation i.e m=n-1), c = no. of
concurrent deviations, 𝑟 also lies between -1 and 1 incl.

REGRESSION

Regression Estimation of one variable for a given value of another variable on the basis of an
Analysis average mathematical relationship between the two variables
Line Regression line of Y on X
Estimation of Y Regression
Regression Coefficient of Y on X denoted by 𝒃𝒚𝒙
(when it is Coefficient
dependent on X) 𝒀 − 𝒀 = 𝒃𝒚𝒙 (𝑿 − 𝑿),
Form
𝑋 𝑎𝑛𝑑 𝑌 𝑎𝑟𝑒 𝑚𝑒𝑎𝑛𝑠 𝑜𝑓 𝑋 𝑠𝑒𝑟𝑖𝑒𝑠 𝑎𝑛𝑑 𝑌 𝑠𝑒𝑟𝑖𝑒𝑠

Line Regression line of X on Y

Estimation of X Regression
Regression Coefficient of X on Y denoted by 𝒃𝒙𝒚
(when it is Coefficient
dependent on Y) 𝑿 − 𝑿 = 𝒃𝒙𝒚 (𝒀 − 𝒀),
Form
𝑋 𝑎𝑛𝑑 𝑌 𝑎𝑟𝑒 𝑚𝑒𝑎𝑛𝑠 𝑜𝑓 𝑋 𝑠𝑒𝑟𝑖𝑒𝑠 𝑎𝑛𝑑 𝑌 𝑠𝑒𝑟𝑖𝑒𝑠

When linear relationship exists The linear equation so arrived can be

between two variables (i.e. correlation used both ways for Y on X and X on Y.
is perfect, 𝑟 = −1 𝑜𝑟 + 1) It means regression lines are identical.
When no linear relationship exist In that case, we need to estimate the
between two variables regression lines with the help of
Important
Theory Points
Method of Least Squares
To derive regression line of y on x The minimisation of vertical distances
in the scatter diagram is to be done
To derive regression line of x on y The minimisation of horizontal
distances in the scatter diagram is to
be done

22 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512
 CA Foundation – STATISTICS | Revision Notes | Important Chapters | Author: CA. Pranav Popat

Covariance between two variables

Defined as the ratio of
Variance of Independent variable
( , )
Regression Regression Coefficient of Y on X 𝑏 = 𝑟. or 𝑏 =
Coefficient
( , )
Regression Coefficient of X on Y 𝑏 = 𝑟. or 𝑏 =

r used here is Karl Pearson’s Correlation Coefficient

Change of origin The regression coefficients remain
unchanged
Change of scale : If original pair is X, Y and modified
pair is U, V where

Properties of
U= and V = , then
Regression lines 𝑝
and coefficient 𝑏 =𝑏 , 𝑏𝑢𝑣 = 𝑏𝑥𝑦
𝑞
Intersection of two regression lines Two regression (if not identical) will
intersect at the point (𝑥̅ , 𝑦) [means]
Relation between correlation and
𝑟 = ± ±𝑏 ×𝑏
regression coefficients
𝑏 ,𝑏 𝑎𝑛𝑑 𝑟 all will have same sign

Coefficient of Determination 𝑟 (square of correlation coefficient)

It explains the percentage of variation
Interpretation of value of 𝑟 in dependent variable due to variation
Coefficient of in independent variable
Determination It means 64% of variation in x is due
Example: if 𝑟 = 0.8, 𝑡ℎ𝑒𝑛 𝑟 = 0.64 to variation in y and remaining 36%
due to other factors. It shows the
reliability of correlation coefficient.

Formula Probable Error [P.E] = 2 3 × Standard Error [S.E.]

1−𝑟
Standard Error
√𝑛
Use Probable Error is used to test the reliability of r
Probable Error
If r is less than PE The value of r is not
significant. Not reliable
Test If r is greater than six The value of r is
times of PE significant and there is
evidence of correlation

23 YouTube: Learn with CA. Pranav, Instagram: @learnwithpranav,

Telegram: @pranavpopat, Twitter: @pranav_2512