:

:*
¥
'

:*
Data :
Data that
collected
are
fact and
figure
summarised
are

analysed and
for
presentation and interpretation .

data
# Statistics is art
of learning from
Descriptive statistics :

et is part of statistics that

is concerned
with description and summarization data
of
ln this we explore data
for purpose of analysis .

Inference statistics :
lt is concerned with conclusion
drawing
from data .

Population and Sample :

Population :

total collection
Population is
of all elements that
interested in is called
we are
population .

Sample :

will be
The of the population that
subgroup
studied in detail is called *ample .

structure and unstructured data :

structured data :
These are data which are
organised
in
predefined fashion
.

Unstructured data :

these not
are data which are

lack data model organised

in
pre-defined fashion or .

Variable and cases :

cases : lt is a unit
for which data
is collected .

Variable : It
is characteristic that
varies across all units .

[ each student )
Eg
: case
Variable C name marks board , , ,
gender :)
Data
Numerical
Categorical
Qualitative variable Quantitative Variable

Describe Continuous
in
float cmeasure) in
integer counts
scale
of measurement ÷

Nominal
} categorical Data
>
.

→ Ordinal .

Interval and
} Numerical Data
→
,

→
Ratio
 Nominal scale :

consist labels
when data is
characteristic
ofobservation or

name
is known
as
as nominal scale
of .

Board GenderokBlood grp etc

Eg Name
: .

. . , ,

Weather →
comfortable uncomfortable
. or .

→
Some nominal variable can be
numerical ( like MIF
yecoaded 011 ) .
=

et doesn't have order

any
→ .

Ordinal scale :

when data exhibits

nominal data but here rank property of are

meaningful this scale

is considered
of
, measurement
ordinal scale as .

Excellent
Eg good hotpoor
: or .

Cold, warm or
.

On this
diff b/w Excellent to
good
→
may
.

not
equal to
diff b/w good to
poor
.
.

Interval scale
of measurement :
alt has all the
data and in this interval
ordinal
b/w values property of
is
expressed in term of a unit
fixed
of measure
is called
than scale
internal scale
. measure
of
.
 Interval data
always
> are numeric .

can
find diff b/w 2 value
.

Temperature 20°C etc

Eg 40°C
→ .

, ,

Here Ratio and you

can't tell 40°C is
has no
meaning
twice hot as 20°C .
,

Ratio scale :
lt has all
property of interval scale
and here ratio
of 2 scalemeasurement
then scale
are

meaningful ,

is called ratio scale

of .

Run etc
Eg Height marks
weight
-
.

, , ,
,

Categorical data :

Frequency distribution
distribution :

Frequency of qualitative
value and their
data
is
listing of distinct
frequency .
 Relative frequency the ratio of frequency :

observation .lt is
to total
of no
generally
-

used to 2 dataset
compare
.

leg
: -

AAABBCCCCD AAAAABCCCCCCDDD

tally Freq Category tally Freq

RF
Category , , RF
A 111 3 •
3
A ☒ 5 0.3

B 11 2 '
2 B I I 0-06

C 1111 4 '

4 C ☒1 6 04

D 1 I •
L D 111 3 02

Total 10 Total 15

Charts
of categorical data :

Basehart and 2 most

piechart are variable
common
display of categorical
.

lnpie chart convert in 300°

frequencyknow share of angle
.

We use
pie Chatto each cat data -
.

Bar
display value
of
horizontal
distinct
qualitative data
with relative
on axis

chart frequency
.

We to
bar each to other
use
category compare
.

When basehart have shorted

frequency
-

it is called pareto chart .

variable is ordinal than chart

If categorical
must be in oodercpaoeto chart)
.

A A
A
B
D D
B B
D C C
C

Pie chart Banhart Pareto chart

 Misleading graph : rotate area
principle .

Area principle that area

by
say
>

occupy
part
amountof graph should correspond
data represented (Diff bar
to
of . .

should not be
of diff width ) .

Trunk at ed
graph ( baseline of graph
→

not with is not practice

start
town hated
zero
brake good .

You may
use
graph by using y-axis
.

%
90 -

A
80 -
80 _

☐
B
70 .
70 .

e
60

s←
.

50 .

(
misleading) Y-axis brake

Measure
it of
of2 centralin tendency :

is
type categorical data :

Median ,

Mode .

Mode
of of categorical column is
: Mode a

most
category
common .

A B Agc A A ,GB Mode A

ModeEgwill have
→
,
B, > =
, , ,

largestwith
bar
largest pie
than
or .

→
If have 2
we

bimodd data category highest value it is

of there is than 2
category than it is multimodel
→
more .
 Median : Median is middle observation of
sorted value 9m ordered form . .

AAA BCABAACCCDABC
Eg
→
"

"
A A A A A A A B B B C C C C D :D
Y ¥ median Codd
median ( even)
)ded
9-1 tries to divide data in 2
half .

Numerical data :

> Desecrate data : 9T is considered as count

of something
.

lf desecrate data is
single value data
→ .

than each value treated as

categories .

Byabout
findingdata
frequency
→
know
.
you can

No
of people diff house 2,43 4,5 2,3 3,3
in .
→ ,
, , , 4,4 1,253,4
,

Converting in category : 2I 32 3 -

.
45

a
f- 5 4 I

5 _
5

4 _
u

3 _
3

2 _
2

I -
I

>
I 2 3 4 5
10 -

20 → 20-10=10 (class width]

East (class marks )
lower
uppddass.es 2012-11
15
=
 >
Continuous data : 9-1 is considered as
measure
something
of .

> there
organise data in number
of
understandable:
classes to make data
Each observation should
belong to
exactly
→

I class .

•
Marks is measured not counted so continues data .

Marks of 20 Student 33,39 30,49 40,30 ÷

,
,

30,40, 30,11 ,
27 , 34,45 48 41,93 , 21,47, 36,33g
, ,

Class interval f- Rt
10 -20 I ◦
0

20-30 3 •
05

30-40 8 •
4
40-50 8 •
4
Total -20 I

continues data is shown

graphical summary
through histogram .

g-

7-

6-

5-

4-

3-

2-

I -

>
% So 30 Go ÉO
Stem -
and-

leaf diagram& : 10th

ltsaperateOn once position
of dictionary smallest to
largest
.

stem
leaf
5
Ig
→
75 = 7
75 , 78 =
7 5,8

Eg 2 15,22 29,3 6,31, 23,45 I 0 25,28, 48

→
, , ,

Steam leaf
I 0 5
2 23 589
3 I 6
4 58

Descriptive measure :
Measures ft indicate
of central
tendency
> :

value centre data set

most
typical or .

→
Measure of dispersion : those measures
indicate variability /Spread in data .

Measures
of central
tendency :

et
capture
> Mean
centre or
typical ness
of dataset .

→ Median
→
Mode
lt is most
commonly used
Mean ÷

lt is
measure .

average means sum

of all
observation divided observation
by
no
of
.
 n -
sample size N →
Population sire

→ mean
reffersas average
.

For desolate observation: -

Sample mean = ☒ = ×, -1×2-1×3 -1 . . .

-1in
n

Population mean -_
µ =
×, -1×2-1×3 -1 . . .
-1in
N

Lecompte → mean
☒
471=7
2 12,5, 7,6 7,3 =
, ,

2,105,5>7,6 7,3 ☒ = 135 19.285

, =

7-
2,105,5>7 6,7 ,

5--1261=21.33
% I. 2 3 45 Total { fixi
"n
% 2 3 54 I 15 Mean __ =✗-
Rift 2 6 15 16 5 44 Ifi
i= ,

4,51=2.93
mean __

Mean
for continuous data :

class interval f- Rf MP timi

10 -20 I ◦
0
15 15
20-30 3 •
05 25 75
30-40 8 •
4 35 280
40-50 8 •
4 45 360
Total -20 g-
Efimi -730
 Frg =

73¥ = 365

→ 36.5
approximation
is not mean be we

are not

midpoint
looking
to data but
. only seeing
add constant to in
→
of you
dataset than every point
old ☒ + constant
new ☒
your
= .

→
of you
dataset
multiply constant to
every
old
point in

than constant ☒ ☒
your
new = * .

outlier
Highly affected by
→
.

Median : -

dataset middle value of dared list

a- is
of a is or .

a- is another
frequently
lt
used measure
divide dataset
of
central
tendency
.

in
top 50% & bottom 50%
-

bn ordered list :( n of observation

•

= no

tf observation is than the odd

data of
no
in Cn 1) 12th value is median + .

if observation is than the

of
no
median is
even

avg of Cnz
and -11 . n
-2
) .

Example →
2,9 ,
4, 6,7 ,
8

On odd .
On even .

arrange data 2,4 6,7 9 2,4 , 6,7 8,9

→
, , ,

" °
median 3rd median
A- 5
52+1 6=2--3 6-2+1--4
= =
=
,

Median 3rd element median

62+1=6.5
-
- =
6 =
 Very loss affected by outlier
.

is added to each
Of dataset
constant
point of
median
,

length
will be
does 'n
old change
median + c.
so new

each data
ofconstant is
multiplied to -

point
-

median
-
Than due to some
will be old
length
median ✗ c.
.

Mode :

value
it is most
frequently occurring
of dataset
occur more than
.

than
of value I
-

no

there is no mode .

I , 2. 3,7 , 7 3,2 , I mode

Eg ÷ 7
=

2,9 , 3,4g 2,7 no mode .

is added to each
Of dataset
constant
point of
mode
,

length
will be
does in
old mode change so
+ c.
new

of constant is multiplied to each data -

point
mode
Than due to some
-

will be oldmode
length ✗ c.
.

outlier
Let's
Don't
affect 2
by
data :
.

compare
D1 3. 3.3.3.3 D2 =L, 2,3,
→
4,5
→

finding DL measure
02
of central
tendency
:

Mean 3 3 However mean ☆ median

Median 3 3
Mode 3 -
are same
dataset
for dataset
are
but
not Pame .
Measure
of Dispersion To describe the
:

above

descriptive
difference
measure
quantaténely
that indicate
,
we use

amount of variation , spread data These

in .

is called measure of dispersion/variance/Spread .

Measure of dispersion are :

→
Range ,

→
Variances
→
standard dauiation ,

Interqiartile range
→
.

Range -9-1 is
defined
: difference as b/w
and lowest value
largest
dataset .
of a

Range Largest lowest

= -

Let's 2 data :
compare
DL 3. 3.3.3.3 D2 =L, 2,3,
→
4,5
Range =3 -3=0
Range 5-1=4
=

103--1,2 3 4,15
, , As we can see it do well in

Range -1-5-1=14 DID but D8 not properly ☐2 in

So ,
Range is extremely enciline to outlier *
.

Variance : On contrast to variance

range
-

takes into account all observations .

to measure
is consider daviatéon variability
dataset
One
way of
-

data value from

centre value of
a- is affected outliers
by
.
 Eg
:
-

12345 Centrepoint =3 ☒ __

-2 -1012 difference from centre

Population variance :

02 (R µ)2+(R2-NY2+ -16cm UP
- -

= , . . .

Sample R variance :

( Rn -502
5- = ( ,
-

Ñ)2+(kz-ÑJ+ .
.
.
+

n -1
(E , -

E) Ck , - IF

Eg DI Deviation from
Data
:
Squared Deviation
-

mean

.
I 68 68-59 =
9 81

2 79 79-59 =
20 400

3 38 38-59 = -21 441

4 657 68--54 = 9 81

5 35 35-59 = -24 578

6 70 70-59 = 11 121

7 61 61-59 = 2 4

8 47 47-59 = -12 144

9 58 58-59 = -1 1
10 66 66-59 =
-7 49
Total 590 0 1898

Population variance -_

18,098--1898
Sample
189*-210.88
variance =

lfa constant isaddedtoalldatapoientthan

the variance doesn't
change
.

lfa constant is multiplied toalldatapoiontthan

new variance oldvarianoe ✗ ( constant )2
-
 Standard deviation
square root
: 9T is
of variance .

Square root of sample variance is sample

standard deviation .

Square root of population variance is

population
-

standard deviation .

Population standard deviation :

0 = Gc, µ)2 + tri NY2+
-
+ ten UP . . .
-

Sample standard deviation :

ÑJ+
2
s = (R ,
-

B) + ( R2 -

.
.
.
+ ( Rn E) 2
-

n
-

Ds .

Population Std deviation

.
=
1898T

Sample Std .
deviation =
210.88

Unit standard deviation

of
:
2
U2 → Variance is recorded unit as
Standard deviation soot convert it back
by
u →

Of constant is added to all data point than

a

the standard deviation doesn't

change than
.

Of a constant is
multiplied to all data point
new 51 deviation old 51 deviation ✗ constant
= .

d- is affected outlier
by
-
.

Percentile Percentile indicates the

distribution
: -

percent
of data
of
.
 To
find percentile :
-

Arrange data in
ascending order
-
.

Divide data into

half if is n even we
-

it is odd don't
get integerthan get
'

if determine
'
we

integer than np Cnp middle of that

smallest
integer
.

part)
greater
=
.

It is value in that position

of loop percentile
-

But if np is
integer and ripof-11
in case even
-
.

avg of rip
than take .

data :
Example : Arranged
35,38
-

,
47 , 58
,
61, 66, 68,68, 70,79

P np
01 I (35+38)/2 =
36.5
025 25 =
47
05 5 (61+66)/2 =
63.5
0.75 75
=
68
I 10 =
79

→
Quartile :

Sample 25th percentile is called 1st

quartile .

The sample 50ᵗʰ percentile is called median/

Second quartile .

The sample 751h percentile is called 3rd

The
quartile .

fine number :

Minimum - summary
1st quartile ( lower )
-

02 03 ( median)
quartile
min a Max
2nd _

median
3rd ( upper )
quartile
-

Maximum
 Interguartibe
→
:

The range
interguartileand
range
LIQR) is
difference b/w first third
quartile .

IQR= Qs -

Qs

Eg
-

1st quartile 01--4975

,

3rd
quartile
(03-01)
,
03--68
IQR 1825 =

lt is also measure
of dispersion .

Contingency table
a- is also called
:

two-way frequency table .is a tabular

mechanism with atleast two row &
two column used in statistics to
present
categoricalintern of count
frequency
.

Eg
:

Gender No Yes Total

Male 10 34 44 nominal data
Female 14
Ownesfshone Total 24
42
76
56
100

Income (coded) Yes No Total

I 2 18 20 Ordinal data
66
By income 2

3
27 39
9 5 14
Total 38 62 100
 >
Row relative frequencies : d- is
dividing
total
each row
by its row .

Gender No Yes
Eg Total
:

Male 101*34/44 44 nominal data

Female 14156
Ownesfshone Total
42/56
24/10076/100
56
100

→ Gender No Yes Total

Male 227k 772% 44
Female 25.0%750%56
Total 240%76.0%100

> Column relative frequencies : d- is

total dividing
each column
by its row .

Gender No Yes Total

Male 10ha 34176 44/100 Nominal data
Female 1424
Ownesfghone Total 24
42176
76
561100
100

Gender No Yes Total

Male 41.6%44.76 46.0%
Female 58.3%552%50006
Total 24 To 100

>
Association b/w two variable
whether
:

A- is
finding
variable
information about
one
provide information about
another variable .
 of the column relative
frequency
→
row or
all than two
fornot
are same rows
variable are associated with each
other .

Gender No Yes Total Both talalob

Row related Male 227k 772% 44 are
for male
Female 25.0%750%56 oft female %
frequency Total 240%76.0%100

Gender No Yes Total Both male

column related Male 41.6%44.76 46.0% &
female %
58.3%55.2%50006 similar
frequency
Female are

Total 24 to 100 to total .

60 ?
no yes
100 -

no yes

40 _
75 .

50 _

20 _

25 _

Female Male Female Male

(Standard bar chart) 11004 stacked bar chart)

→ Stacked bar chart :

-
lt summarise data in
with
form of banhart
proportion in
respect to category .
 of the row or column relative
frequency
>

rows than two

are different
variable for some are associated with each other .

Income level Yes No total

there ,
in both High 10.0%90.0%20
route column Medium 409459.1% 66
low 641%35-96
freq totaldiff are 14
.
.

38.0%62-0%100
from income
So ,
Total

* phone Income level Yes No total

are connected .

High 52 % 290%2006
Medium 71.0% 62.94 66.06
low 23-6 % 81 % 14.08
Total 38 62 100

that
Scatter plot a- is
graph displays
: a

pair of values as
points on 2- ☐ plane .

7-
Age Height
I 75
←

100 - '
'

'

2 85
.
* .

60

3 94
_

↳ 101 20 -

5 1 08 ' '
%
"

¥
'

%
•

e 2 y s

You can describe association b/w variables

in scatter plot
by answering 4
question :

>
Direction → Does pattern Up Down
,
or both .

>

,
 → Cavaliere > bit t.in ear or curve .

'
'

tighlydiiiÉÉed iispaceblw points

. .

> Variation > .

:
ii.÷:: : ::
:
: :
::
> Outliers > Outside pattern (exceptions) .

Measure of association :

strength of association b/w 2 variable can be

measured with : -

Covariance
-

Correlation

quantify strength of linear

covariance et :
-

relation b/w 2numer.ec variables ( . No units )

Hi I) -

( Yi 5) -

Age Height
I 75
Deviation
-2
✗ Deviation
-17.6
( a-
35.2
ly , -5)

7- 6
2 85
-

I -7.6
3 94 0 1-4 0 I

↳ 101 I 8.4 8.4

5 108 2 15.4 30.8
☒ =3 5=92.6 5=82

Age Height
I 6
Deviation
-2
✗ Deviation 's
2
( a- Cy , -5)
-

y
2 5 -1 I -1

3 4 O O O 2

↳ 3 I -

I -

I
5 2
2 -2 -4
☒ =3 5=4 4=40
 2
' ' mean
Cov > 0 Corso \
Y Y
° ki E)70
Lying > ^
-
'
-

( Ri ÑKO
-
cyi.gl> 0
DO _
6- .

'

100 _ .
5- .
.

'
80 _
4 _
.
.

I -

60 3
_
_
.

( Rix )
40 Yi F) <
-° 2- .
(y g)< o
-

,
-

20 _
( Rix ) >0 I _
.

" " " '

¥
" • " '

¥
'

%
' •

%
•

I 2 3 y 5 I 2 3 y 5

associated with
when
large ✗
large 's and
>
is
vice-versa the deviation
sign will be
associated with small Y and
same
When
large ✗
>
is
vice-versa the deviation
sign will different . .

The covariance b/w variable ✗ try is

given by :-

Population covariance
n
:

Cr Illy g)
-
-

Cover g) ,
=
i=1
N

sample covariance
n
:

Cr Illy g)
-
-

Cover g) ,
=
i=1
n -1

1
Population covariance
-8,1=164
- 2
Pop -
Cov .
=
=
-2

Sample covariance
-841=205
-

Sam .cov=
-41=-2-5
 Correlation A- is more
interpreted easily
:

linear association b/w

measure
of
two numerical variable ( No units ) .

d- is derived
from covariance
→ .

correlation b/w two variable ✗ by

→
To find
divide covariance btw ✗ 44
byproduct
of standard deviation of ✗& .

> Correlation = Covariance

standard Deviation

( r F) ( ,
-

, -5)
> Correlation =
it
=
COV ( R , Y)
"
( Ri -
Ei
n
( y ,→, Sxsy
E- I E- 1

Hi -732 , -55
Age Height
I 75
sq.Dev.ofxsq.Dev.ofxlri-xxy.FI
-
E- 4 -17.6=30976 35.2

2
" I -7.5=5776 76
85
-
=

= 0
3 94 0
1-42 1.96 :

4 101 [ = '
8.4%7056 8.4
5 108 Ñ= 4 15.4=23716 30.8
☒ =3 5=926 Elo E- 6772 5=82

512--1.58 , SDY -1301 ,

covariance 205

Correlation __
Cov Cri) =82_ Or 2 0.9964
Sxsy THE 1.58×13-07
 Hi -732 -55
Age Height
1 6
-
sq.Dev.ofxsq.Dev.ofxcri-xxy.FI
I y 22 = 4
y
= -

2 5 -5 =
I 5 = I -1

3 4 =
0 = 0 0

↳ 3 i =
i -

i = I -

I
5 2 22 =
& -22--4 -4
☒ =3 5=4 2=10 EIO E. = -10

512=1.58 , SDy= 158

,
covariance

correlation __
Cover,Y) =
-10 or -2.5 =
-
&
Sxsy Foxx 1-58×158
2
mean
Corr Correo
' '
>0 1
Y Y
° ki ñ )> 0
Lying) > ^
-
'
-

( Ri ñK0
-
cyi.gl> 0
DO _
6- .

'

100 _ .
5- .
.

-
80 _
4 _
.
,

I -

60 3
_
_
.

( Rix )
40 Yi -81<0 2- .
(y g)< o
-

,
-

20 _
( Rix ) >0 I _
.

" " " '

¥
" " ' '

¥
'

%
' •

%
•

I 2 3 y 5 I 2 3 y 5

These linear relation can be summarised

through line .
Point Bi serial
-

correlation coefficient :

> there data (one numerical

group
wee our
column
and one
categorical and encode
female=D
our
categorical column
eg
→ male -0
-

.
,

value numerical
→
compute
column
mean
of
in respect to encoded cat column .
.

male/Female student
egPo
→ mean
of marks of .

→
and Ps is proportion of group Eg . :
-

For 20 student male =

12120 ,
female-8120
So Correlation
,
Coffecient :

Jo -51
Tpb =
Pops
Sx
 PERMUTATION & COMBINATION
Adding rule of counting :

9f action 1- can occur in a different

another action B ways
than
,

can occur na diff ways

.

total number of action A or B is Ms

occuranca
4 shirt and
of have to
+ Ns .

Eg
: There are 5 paints and we

select either 1 shirt or I paint than no

'

of
possiblity to choose will be 5+4 10
ways
=
.

Multiplicate rule of counting :

action A can occur in a different

9f
another action B can occur na diff ways
than
,

ways
.

total number of occurrence action A and B is Ms ✗ Ms

of .

Suppose that i action are there than there are

'

Ni + Nat + Nr r'actions
possibility altogether for
'

. . .
.

There 4 shirt and 5 paints and have to

Eg are
: we

select either 1 shirt & I paint than no

'

of
possiblity to choose will be 5×4--20 ways
.
.

add 2 color shoe than it will 5×4×2=40

lf we .

Eg : Your task is to calculate how

many ways create to
6
digit alpha numeric
-

password with first 2

digit alphabet and last 4 digit
as as numbers .

If reputation allowed : alphabet -26 ways -

numbers -10 -

ways
.

ways
=
26×26×10 ✗ 10X 10 =
6760000 .

not allowed
If reputation : 2625-10987
=
26×25×10×9 ✗ 8×7--3276000
Eg → There alhokte who take part in 100m race
What are possible
are

athlete can finish race

8 .

ways
.

→ Position I 2 3 4 5 6 7 8
8 7 6 5 4 32 4
ways 8×7×6×5 ✗ 4×3×2×4=40.320
No
of ways
=

Factorial The product of first& n'

veintkounting
'
: +

number )
n ! . n ! =
is
n ✗
called
(n -
1) ×
factorial
.1 . . ☒
denoted
by
Note : 0 ! =L

So , above example can solved as : 8! =

40.320 .

It also written as i 5 ! = 5×4 ! =

5×4×3 ! & so on .

So i≤
9h
general for have ,
n
, we
,

! n ( = n ✗ n -

1) • a. ✗ ( n it 1) ✗ ( n
-
-
i) !

Eg :

6g,÷ =
6×5×4113×24
3×2×5
=
6×5×4--120 Or 6×5×4×34
34

6! ✗ 51
=
6×5×4×3 ! ✗
5×4 ! =
6×5×4×5=600
31×4 ! 3! 4!

Eg :

Write 25×24×23 in term of factorials :

-

25×241123×22×21 ✗ . . # 25 !
=
=

22×21 ✗ . . . ✗ I 22 !

Permutation : 9-1 is an ordained

arrangement of
all of object
'

or some n' .

The number of possible permutation of

 N =
Objects
8 =
positions

objects from collection of districts is

given
' '
'
o a n'

by formula
:

N ✗ (n - 1) ✗ . .
.
✗ (n - 81-1 )
& is denoted "

by Pr

MPs = N ✗ (n -
1) ✗ ( n 2) ✗ -

. .
✗ ( n -81-1 )
^
Pr = n ✗ (n -

1) ✗ ( n 2)✗
-
. . . ✗ In -8-1 1) ✗ ( n 8)( n
-
-
8- 1)* . .
✗ I

( n 8)( n
-
-
8- 1)× . .
✗ I

npr =
n !
a- 8) !

Special cases :

1.) npo = n ! ni I
n;-
= =

(n -031 .

2.) MPs = n !
= n

( n -
y!
3.) npn = I n ' n! where 0 ! = 1 .

G-
=

(n ng ! -

Eg : No
of ways of 8 person committee can be choosers
for
Chairman & vice chairman without repititon .

n -8 8! 56
ways
-

=
.

22 (8-2) !
No of 4
digit no can be formed using digits
1,2 , 3, 4,5 if no
digit is repeated .

>
=
n -5 -

=
5! =
120
8--4 1 !
 How no can be
formed ? Same
many
even .

- _
2-
- -
-
-4 fixed
Sog N = 4
g
8--3

for for both 24×2=48

4¥
=
=
24 I do .

6 people chair in linear order ?

10
Ways any
.

N = 10 ,
8--6
'°
Po =
151200

3 characters A. B. c can repeat how

many ways
3 box ?
to fill
3W 3W Sways 3×3×3--27
ways
=

same A. B. C with 2 box ? 33--27

3W 3W =
3×3--9 32--9
ways
The possible permutation of
of objects
' '

no
.
r

collection of an distinct object where

from
'
a

nepition is allowed is given by nxnx xD . . . .

.

denoted
by no
.

Example of rearranging objects :

how ?
DATA times it
many
can be
rearranged
DATA
=
DATA DANA 198 ,
, we consider A different )
AiTAzDAaTA.D (
ATAD
similarly )
 4! if total ways to
arrange
4 letter wordsin .

2! is place which 2 a' in data

' '
'
2 are same =

'

/2 !
arranged
'

DATA be in 4 ! 12
ways
=
can .

No of permutation of object when p of them n

are
of I kind and rest distinct is equal to n !
P!

No of permutation
-

of
'

n'
object where ps is one kind &
Pa is diff .

Kind and so on so , eq will be

.
n!

17 ! Da ! . . .
Pn !
For statistics
' '

n =
to p=5 =
10 ! =
50 ↳ ◦ ° >

s
=
3 3! ✗3 ! ✗ 11×2! ✗ 1 !
it =3

a
=
I

i = 2
C =
I

circular
arrangement
:

A C

rotated
D B = B D Both are same
just .

C
A

A A

D B ≠ B D

C C
Mirror
images are
different .

anticlockwise rotation
9f
than
clockwise
will
or

be a- 1) !
are some

it
2
 anticlockwise rotation different
9f clockwise or are

point will be Cn 1) !
means
if we
fix one
-

A C

D B ≠ B D

C
A

Ex Solve
for
. n .

"
Pu =
20hPa =
A- =
20 ✗ MI
En -41 ! In -27 !
( n 2) -
✗ ( n 3) -
=
20 do , D= -2 or
n=7✓
,

1%-3
T.hn?En:?; ---3-Bnn--y
=

§
= → 2h
= 20 B n -
-

to

solve foot LHS RHS

•
513=2 Pot 5ps 5! 6%-1
,y=§
. =
=

(5-8) !
2×61
↳
¥÷y .
=

(7-8) ! :
-

(7-8) ! =
G- 8) ( 6- 8) (5-8) !

É¥=¥→%?¥→ → (7-8116-8)=12
82-130+30--0 0--3008--10
Combination:
In combination ab=ba
is
The electing possibleofation daterobject
number comber

by "C and is
given by: wer=u!)!
R!

Some useful resel.

The
v
[ cn-r
ran-r!-ta!r! "

In other word,
selecting object from 'n' objects
is same as
rejecting new objects from' objects
"In and "Co:1 for all values
1
of
=
n

Crest "Cr;
4-1
"Cr =
KUKR

Examples en with
paper is divided part
exam a in a
A student need to attend
↳95
question.
8 ques & min. of I have to done in both.
* TC3s+'cy"CutC, 3Cs=35+1757218-420
ways.
Total of ways to choose card 52 cards?
52Cn
no

526
from
270725
=> =

↳! 48!
All" card of same suits?
48
x32n=uxted:28
 Select 11
there and
players from
bowler
17
player where
must be their
5 bowler
?
are

4
Ways
•

5cg ✗ 12
Cy 5!
=

×
12 !
=
5×950402=3960
4! 1 ! 7!5 ! 120
ways
In points of
that
general given
be drawn
in
, ,
number
point
line
segment
can
connecting is ncz .

Distinguish situation btw permutation & combination :

→
Permutation is used when Onder matters .

→
Combination is used when order don't matter .

Eg
:

Perm : No
of ways 8 athlete can come 1st 2nd 43rd
.

athlete
Comb : of ways qualify
No 3
from 8 -

Comb : Choose 2 chaos representative from 40 student .

Perm : choose I v. c & 1C from 40 students .

Similarly going
lines
Direction matters
to dot if :
Permutation - .

Direction don't matter → Combination .

Probability :

Experiment 9T is: that produce

any process an

an observation or outcome .
 Random experiment : lt is experiment whose
an

outcome is not predictable with certainty .

Sample space Cross) 9T collection basic outcomes

: is
of all .

Ex : Coin toss ( HT) ,

Die roll ( 1. 2.3.4.5 b) .

Event : it is collection
of basic outcomes . 9T is subset of
*
ample space We event occurred if event is
say
.

contained in subset
Eg : Die roll ( 1.3.63 coin toss ( HH)
.
.

Union of event ( U) : 2 events A and B ,

we
define
event AUB called union
new
of event Ah B. to
all outcomes that are in A&B or in both .

AUB occur if either A or B occur .

Intersection event
of event ( n ) : For 2 or more we

define An B as intersection of event if all

outcomes are in both A & B.
AUB = ∅

Venn
Disjoint
diagram event

ACB

A
☒

Subset
Null event and disjoint event :
Event without outcomes are called null event
any
.

( Ol )
Symbol
If intersection of A and B is null event than
A & B can't occur
together do A and B is
( An 13=0 )
disjoint
00
mutually exclusive event .

compliment of event an :

all
'

A compliment of A is denoted
by A means
outcome of sample spaces that are not in A.

Eg The
Coin toss
: 5- { Hit HT.TT TH} A {HT} A ÷ { HH.TT TH }
=
.
, , ,

complement of sample set is will set 0 .

If all outcome eventsof are in event B than we can

A is contained B A is subset B- A CB
say by or
of
Eg 2 coin toss S { HH HT, TT , TH }
' =
.
,

A Ist = head =
{ HH , HT} ,
Both head =
{ HM } B =

we can BCA
say

ProbabilityClassical :

bets be sample space

def :
of
random exp in which n'
likely outcome equally
'
.

& E event consist m of these outcome the

' '

event poobility is min & PCE) m

-

every
-

ñ
Even dice roll
Eg
: in -

no

6=(1,2/3,415,6)
3=12,4 b)
My 62 1-2
m = n = =
, = =
 Relative 9T calculated
frequency is
by repeating
:

experiments times other words if

ln
many
,
.

(E) is occurs in ñ repetition

n no
of times E

of experiment .
PC E) =

Giz %Eˢ

Eg 9h 1000 toss
you get
head 468 times
:
so

probability of head PCH )

4¥
468 times
= ◦

subjective approch This probability of event is best :

statement This
guess by person making
.

individual's
probability is measured by
degree of belief in event .

There tomorrow
Eg % chance is 70
you
will win
: .

Probability
For
Axioms
event E. the probability of E
:

is number
any
between 0 and I That is 0 ≤ PCE ) ≤ I .
, .

The
probability of sample space is 1. ( the outcome
with PCE ]=D
of exp
.
will be element
of sample spaces
The
probability of the union of disjoint event is equal
to sum of
prob of these events PCE UEa) =P(E) + RED . .
,

;)
•

P JE =
Pl Ei )
i I -
- E- 1

Dice roll { 3.5) Eleven) {2,46 }

Eg odd no )
E. ( 1,
Getting
: .
= =

E,
3g Es
%
=
=

PIE U ED
} +3g 8- I
=
, =
=
General property of probability :

Probability of compliment of an event :

E- and E ' is
disjoint EUEC -5
-

PCS)=I PCE ) ,
-1 PCE )= I
'

PCE ] =L PCE)
'
-

P( 01--0
Probablity
5=0
-
of null event

et also work
-

Addition rule of probability :

for not
-

disjoint
PCE.VE ) =P (E.) PIED . + -

P( Eines)

a •oB

E. UEz= IUIIUIII
I II III E ,
= IUII

Ez = IIUIII
Et Ez
• Mi Ein Ez = TI

PCE,UEz)=P( IUIIUIII) > PCI)tP(E) + PCII ) -

①
PCE ) ,
PLIUII )
=
> PCI )+P(II) -

①
P( Ez) =P # VIII ) > PCI)tP(TI) -

④
PIE,nE2 ) =P ( TI ) > PHI ) -

From 11,1114W
get
1 we :
,

PCE.VE ) =P (G) + PLED

.
-

P( Eines)
Eg→ both
Customer probablity to
buy shirt 3 apart neither -2 •

1. what is •
is probability of customer to buy .

5=-3 ,
P= •
2 ,
Snp -_ •
1 then hither will be P (SUPT
:-P ( sup)= Pls] -1 PCP] PCPAS ) -
= .
3-1.2 -
•
I = •
4
thence PCSUP ) ? 1- 4
,
6 ◦
= •

Own smartphone
Gender No Yes Rowt .

Female 10 32 44
Male 14 42 56

Column 1- 24 76 100

Own smartphone Being female not

Rowt
Gender owning smartphone
No Yes .

Female •

10 •

32 •

44
Male ,y 42 56 Probof being male
•
. •
.

column 1- 76
owning smartphone
•

24 • 100 ☆ .

Joint
Probablityigt
intersection
represent probablity
events
of
2 of or more
.

Eg : Female & not

owning smartphone = -10

Marginal probability :9t is

probability of observing an outcome
with
single
attribute
regardless of its other attributes .

Pl PCFO )tP(FO )
'

Lg : female) 10-1.34=-44
•
=
=

Planning Phone )= •
34-1.42=70
Conditional
probability
:

Finding condo prob .

to answer
question like :

>
Among femalenotbuyers what chance
of owningmale
phone
' '
=
is

Among people have phone ?

'
how are
many
,

In this restrict the sample space

table to
we

a row or column .
of contingency
Gender No Yes Total
Male
Female
10kt
14156
34/44
42/56
44
56
By Row
Total 24/10076/100 100

Gender No Yes Total

Column
Male
Female
10ha
14124
34176
42/76
441100
561100
By
Total 24 76 100

PCEIF ) d- is conditional
probability of E-
given
-

f- has occurred .

The of that
probability event E occurs
given
event

given by
:
f- occurs .is

PCE / F) = PCENF ) ; P( F) O E F
P( f)
5
 Multiplication rule : conditional
probability formula
:

Multiplying both side with PCM

get , we :

P (f) .

PCE I F) =
P ( EAF ) . P (f)
P (F)
•

: PCE n F) = P (f) •

PCE / F)
9T states that prob of .
both E &
Foaming together
is
equal to probability that F occurs multiplied by
cond prob of E F occurs
given
.
. .

lt is
useful to compute proto of intersection of event

Eg :

Cond I
Of 40
-
student male 17
female
23
list st female & 2nd male )
-
.

P ( f, n Ms) =
PCF ) ,
•
P ( Mal F.)

% 233-9 251
= •
• =

This rule can be

generalised intersection of an

event
arbitrary no
of is robbers as :

PCE MEN . . . A En ) =
PLEA P( Eat E) PIE 31 Ein Ez ) . . . P (Ent Ein Ez . .
.

n En -
s )

Eg : A deck of
the
52 card .
divided in 4 pile randomly .

What
Ace of
is
probability of following events .

Er :
spade is in
any of pile 4 .

PLE ) ,
= I

Es : Ace of spade & Ace of heart in diff pile .

-
 On 1st pile Ace of Spade of 13 Card is card
1
card will choosers card so
rest 12 be from 51 .

Ans will be its complement PEI ED-1 -1¥ .

=
3¥
E3 : AS AH AD . , in diff .

pile .

Pi and AM in P2 so to
we put As in
fillSorest
12-12=24 card A & As we have 50 cards
of no
'

. ,

A & Pa complement
of ways AD can be in is

PLEW EA Ez)= I
of -7¥ =

3¥
Pl Eal Ein Ean Es )
got by fixing AS AH
we

Ey : =
12 (
Firing 3 card in and3 deeds
49
choosing from 36149 complement
Pl Ein Ean Es MEU)=
3¥ 3¥ ¥9
:
Eg .
.
≈
0105

Independent event 9T be defined the

upcoming
:
can as

events doesn't depend on

previous event
.

coin toss ( HT ) don't depend on

prev. result
: Current

Eg
Since :
.

PCE n F) =p (F) ✗ PIE I F)

Hence Two event E and F are independent if
,

PCENFJ =P (F) XP (E)

The that both equals to the
probabilitytheir occur

product of individual
probability -

Dependent event : Two event that are not independent

is dependent .
Eg Rolling
: 2 dice .

§
( 1 1) ( 1 2) ( 1 3) ( 1 4) ( 1 5) ( 1 6)
, , , , , , , , , , , ,

( 2 1) ( 2 2) ( 2,3 ) ( 2 4) ( 2 5) ( 2 6)
, , , , , , , , , , ,

( 3 1) ( 3 2) ( 3,3 ) ( 3 4) ( 3 5) ( 3 6)
, , , , , , , , , , ,

g =
,
,, ,
( 4 , , ( 4,3 , ( g µ , ( g g, , g g ,
, , , , , , , , ,

(5 ,
1) ,
( 5 2) ( 5,3 ) ( 5 4) ( 5 5) ( 5 6)
, , , , , , , , ,

g, , , ,
, ,,, , ,
, ,
, ,, , , ,
, , ,
, , , ,, ,
,,

Define foll
Ed
.

:
events :
First outcome is 3
6136 =
1/6

Ez : Sum
of outcome is 8 .

5136

E3 : sum outcome
of
6136 =
116
is 7 .

AreEin
event
Ea
E. & Ea independent ? No
→ is 1st outcome 3 & sum of outcome is 8 .

Only event is ( 3,5) is 436

Since ,
436 ≠ 6136×5/36 we see ,

PCE NED ≠
,
PC E.) ✗ PIEa) , so E. & Es are not independent .

Are event E. & Ers independent ? Yes

Since , Ein Ez =
P (3/5) =
1136 .

Since PCE in E3) =P (E) ✗ PLED

1136 6136×6136 do it is
=

independent .
 '

Independence of E and F
E and F independent does E and F?
If are so

'
En F
E En F F

En F & EUF is
disjoint mutually& exclusive
'
.

'

So , P (E) PCEN F) +
=
PCEAF )
PCE ) =P(E) ✗ PlF) + PLEA F)
PCE ) P (E) xp (F)
-

PCENF )
' =

PCENF c) =P (E) [ I P (f) ] -

'
PCENFY =P (E) Ptfe) so , Elif are independent .

Independent of 3 events :

3 events E. F and G are said to be independent

if i PCENFA G) =P (E) ✗ P (f) XP (G)
P ( En F) PCE ) ✗ PC F)
=

Pl Ena ) PC E) ✗ PC 6)
=

Pl Fn G) PC F) ✗ PCG)
=

9f all these 4
happens than it is independent event .

Law of total
probability
: lot E and F be events In order
to outcome to be E it must be in both E& For bein E and
.

not in F.
'
Ehf
E En F F E = ( En F) UCEMF c)
 Formula & interpretation :
PCE ) =
PCE NF ) + PCEAFC)
=P (f) PCEIF ) + PCF c) PCE IF c)

> E

Fi FI F3 Fk -

E =
( En FDU ( En Fa) UCENF}) U ( ENE )
P (E) PLEA f.) + p( Enfz) -1 P( Enfz ) + P ( En Fk )
=

P (E) PCE / f.) Plf )tP( Elf )P(E) + PCE/Fs ) P (B) +

= •
,
,
. . . .
+ PCE/ Fr ) PCFK)

suppose event Fi f. ifs & Fr are

, .
mutually exclusive
exhaustive ; that is the event must occur
&
exactly
one of
.

Than event E
for any
-

P (E) =
PCE / f.) Plf ) -1 PIE / f) PCE ) + PCE If3) P (B) +
•
, • . . .
+ PCE/ Fr ) PCFK)

"

p( Ei ) =
PIE / Fi )P( Fi )
E- I

Bayes rule Suppose interested conditional

'
: we are in

prob of .
event F conditioned on E- we know

PCF / E) =
PC Fn E)
P (E)
 From def:
PCFIE) =
PIFTE) -
PCEIF) PCF)
PCES ↑CF) PLEIF) + DCFY) PLEIF4

Bayes' rule exclusive

is
suppose that event F1,2,
and exhaustive; than
FK, ..., are

mutually for
event E.
any A CFIE)= PCEIF) PCF)
K
↑ (EIFi) P(Fi)
=I

Sample space: A
distribution
random variable is numerical
of outcome of statistical experiment.

Eg: Rolling & dice.

S S
CL, 1), (1,2), (1,3), (1,4), (1,5), (1,6),
( 2 1), (2,2), (2,3), (2, 4), (2, 5), (2,6),

33;·13.3,23,)3C,Cs
1 ), ), 2), 33), ( ·U), (, 5), 2,6),

S -

(
3 3 3 is3 33.6
3
13.),2),)s3, he

26, 1), (6.2), (6,3), (6,4), (6, 5), (6,6),

of the outcomes, how

many
outcomes
outcomes will result
in sum
of 7? as

of how
many
outcome will have smaller of the
outcome is 3?

(sample space for both question is same)

Outcome ✗ Outcome ✗ Outcome ✗

( 1,1 ) 2 1
( 1,3 ) 4 1 ( 1,5) 6 1
( 2,1 ) 3 1 ( 2,3 ) 5 2 ( 2,5) 7
2

( 3,1 ) 4 1 ( 3,3 ) 6 3 ( 3,5) 8 3

( 4,1 ) 5 1
( 4,3 ) 7 3 ( 4,5) 9 4
(5,1 ) 6 1 ( 5,3 ) 8 3 ( 5,5) 10 5
( 6,1 ) 7 1
( 6,3 ) 9 3 ( 6,5) 11 5

( 1,2 ) 3 1
5 1 ( 1,6 ) 7 1
( 1,4 )
( 2,2 ) 4 2 ( 2,4 ) 6 2 ( 2,6 ) 8 2

( 3,2 ) 5 2 ( 3,4 ) 7 3 ( 3,6 ) 9 3

( 4,2 ) 6 2
( 4,4 ) 8 4 ( 4,6 ) 10 4

( 5,2 ) 7 2 ( 5,4 ) 9 4 ( 5,6 ) 11 5

( 6,2 ) 8 2 ( 6,4 ) 10 4 ( 6,6 ) 12 6

bet ✗ denote sum of outcome of 2 rolls .

✗ takes the value 2.3.45 6,718,9 10,11412 , ,

.

Relevant
✗ Value
{(
event
Probablitylk )
2 1,1 ) } V36
3 { ( 12112,1 ) } 2136
4 { (1,31122%13,1) } 3136
'
h O i r
G r - -

10 {14,6%(5,5%16,4)} 3136
11 { (5,6%6,5) } 2136
12 { ( 6,6 ) } 1136

bet Y denote lesser outcomes

one
of 2 .

Y takes value 1,234,546

 Relevant
Yvalwe
{(1,1%11,2)
event
Probablitylk )
I ,
-
→ (5,1%(6/1))=11 11/36
2 -
-
-

=9 9136
3
- .
.

=7 7/36
↳
. . -

=5 5136
5 { (5,5%15,6%6,5) } =3 3136
6 { ( 6,6 ) } = I 1136
Answer
Eg :3 coin
S={
are tossed .

HHH.HHT.HTH.HTT.THH.TN/-eTTHsTTT}
are head ?
Of these 3. tosses How
many
"
" "
"
which result in head first time inpos ?
.

→ bet ✗ =

hoof head ,
✗ take value 0.1.2.3

Value
of ✗ Relevant event Probablitycx)
0 { (TTT) } 118
I {( HTT),(THT) > ( TTM)} 318
2 { ( HHT) , ( HTM) ,( TTH)} 318
3 { IHHH) }
'
13

g- Toss with 1st time head ,

Y lakes value 12,3

Value
of ✗ Relevant event Probablitycx)
I {( HHH ),( HHTI.LHTHJ.CH/-TD 418
2 {CTHH)g( THT )} 218
3 {LTTH ) } 118
Nil { ( TTT)} 48
Eg: life insurance
agent has
death.
a
elderly client who are

paid th
upon

Let A be event to die and

>
for younger among
I be event for older among
2 to diz

Next year
mayhappen
Conditions t
Money
Both so *
IB 2C
Only younger die A AeBa I (

Only
Both
die i
elder
survive
A? B
B
IC
8
Evalues
Assume A&B are
independent event with
probabilities PCA) == 05 & PIB =.10

let x is total to their

money company pay
client in case
of death (units in Lakhs.
So, X can take 3 values, O, 1 and 2.

Relevent event
value of X

"BC
* PCAY X PCBY 95 x.9 0855
= =
=

d CAR BY UCAMB) PLARB4 +P(A4B) 605x. 9+1 95x=.

=
C 14
& AB P CA3 xP(B) =- 05X.1==005

continous random variable: When outcome

for random
event are numeric a can't be counted a are in
finitely
devisible. Eg: Temperature of person, Height of person, etc.
it involve
typically measuring.
Discrete random variable: It is a random variable
that can take on at most a countable number of
possible value. Eg: No of accident, no of people, etc.
involve
counting on no line.

Probablity
In descrete
mass
function:
random variable. We take countable
value
no
of possible
So let X discrete random variable with
be a
is possible value. Then it can lebled as M., 232,.. n.
por descrete random variable 1, we define
probablity mass
for CPMF) Pins of x by
↑(Ri)=DCX= i)
Re
1x=Ui)
TE, ...
En

* P(U) Pluc]... (Pun)

The PMF Plus is the
for atmost a countable no of
value of X. Thatis x must assume one of the
values x1,X2, . . .

, then
1) (vi)-O, i 1.2, = ...

P(ii)=0 for all other value of <

Represented in above tobular form.

Since x must take one of the value Xi, we have
a

PCR) = 1
== 1

Example: Suppose X is a random variable which takes

3 values 0,192 with probability:
 * PCO) PCI) P12)
↑
(x=xi) // 4 1/2 1/2

A condition is all should be more than ou

· Sum should be 1.14+ 1/2+14=& I

Suppose X is a random variable takes value

0, 1.2, ... with probablities.
P1)
co, for some
= the X

Eg: Rolling & dice.

S S
CL, 1), (1,2), (1,3), (1,4), (1,5), (1,6),
( 2 1), (2,2), (2,3), (2, 4), (2, 5), (2,6),

33;·13.3,23,),3C,Cs
1 ), ), 2), 33), ( ·U), (, 5), 2,6),

S=
(
3 3 3 33 33?;6
3
13.),2),)s3, he

26, 1), (6.2), (6,3), (6,4), (6, 5), (6,6),

All roll map to its sum: like (1,1) 2
4 (6, 6) 12 = =

So, X is a random variable to sum outcomes

maps of
2
X 3 4 5 6 7 8 9 10 11 12
-

(x=1)
↑

if ff f gs t
verify P(xi)=8=I
To

I 1
=

a) lesser value random variable X I 2 B 4 5 b

ply:yis t 85
Sum:1 so, it is PMF.
Graph of probability mass function:
helpful to
It illustrate
is mass
for in a
graphical
0x-xi) the
format by plotting on y-axis
against
xi theon x-axis.

of previous example:
·
 Comulative distribution
function:
Flal=P(x1a).
It can be
expressed by
possible
who
9fy a variable
descreterandom
is

rests... at
distribution for For X is a step function.
It will be step to. So for an
example.

Eg
·
We want to
analyse no
of credit card owned

people.
byRandom experiment:
Random variable: Selecting
No of
adults at random.
cc owned
by person

Probablty
Eg
: consider
-

Even
a
game of rolling dice
outcome You loose amt
- .
.

ondice.

Odd outcome You win amt on dice

-
.
-
Would
play
:

you

hundred
winning always
will after many in ve
-

rolls so won't
we
play outcome is Ho
.

Relative each
frequency of
.

will be
So ,
gain If -2 f- +3£ - ↳
f- -1-501--6f- =
-0.5

Expectation of random variable :

Let be desolate random variable

✗

values Expected value of ✗ denoted

taking EG)
×, ✗a ,
by
-
. .

oeffers as expectation of ✗
given by
:

E- (X ) = ✗i P (✗ =
✗i )
i =L

considered obs
9T is as
long run
avg of
ind
.
. . .
 random variable:
Expectation of a
fun of a

it is expected value of probability by for go

Suppose: X I
X Xe:.. XU

g(d) 9 (K.) g(kz]... 9(Un)

Pu n) =
Ph=niPenine) D(=nn)

expected value
giveg(x) of =

gxPCX=xi)
if a and
are constayEx) at +6

Eg:

Expectation of sum
of a random variable:
->
expected value of sum of
random variable is
same individual random values.
as sum
of
E(x+1)= ECX) + ECY)

Eg:
 expectation of sum
of
and Var
many
expected value of sum of random variable is
same individual random values.
as sum
Let, X., Xe,...,
of
XK be K descrete rand. Var.. Then,

Exi =

Ei
Example: Tossing a coin three times
Toss a coin i times.
Let Xi be a random variable which equals 1 if the outcome is
a head, 0 otherwise.
E(Xi)=0.5
X1 +X2 +...+Xn is the total number of heads in n tosses of the coin.
E(X1 +X2 +...+Xn)=XE(Xi)=0.5×n
For n=3,X1+X2+X3 is equal to the number of heads in three tosses of a coin.
E(X1 +X2 +X3) = 3×0.5 = 1.5
This is the same expectation of number of heads in three tosses of a coin.

The expected of a random variable gives

value
the
weighted average of possible values of
random variable. It don't tell about variation & spread.
Variance
of a random variable:
let expected
value of a random variable x be M.
let be rand Var with expected value M, than the
variance of X, denoted Var (x) or VIX),
by defined by
Var (x) Ex-4j =

x measures the square of diff of rand. Var

its M, on the
from average.
 Var (x) =
ECX-M4
(x-1)
2
=
X2-2XM+M2
Using expectation property:
ECX= 2XMMY =
ECXY-2ME1X+ M same as
var (X)= ECX3-M2

Parnauli Random Variable:

The Rand. Var that takes either 0 or 1 is called
Barnauli Random Variable.
Descrete uniform Random variable:
let x value 11,
be a rand-var
likely to take
any
. . .
.

PMF:

Variance of
let x be a
a
fan of random
random variable,
variable:
let a be a constant,
than, Var (CX) c"Var(X)
=

Var(x +2) Var (X)

=

So, if a bare constant,

v Cax+b) = a VarCX)

Proof:
Ecaxtb)=aM+b,
VarCax-b) Eaxtb-aN-b=ECarx-4/2)
=

=
aLE (X-MP=a " Var (x)

Variance
of of
a random variable:
sum

We know, ECX+YS:ECE) + ECY)

let x 44 are I random variable
Let x= x &
y=X,80(v= variance)
vx+Y)= vx+x) VC2K) 4VCX) = =

So, VaR(x+ x) Var (2X)= 4Var(X) E Var(x) + Var (X)

=
 Independendent random variable:
X Y are independent R.V if knowing value
of one
of them don't
change prob of other.
Ex. -Roll a dice twice. S=3(1,1), . . .
. (6, 6) 3
X
->

& x outcome of 1st & 2nd dice

x=i don't value 12,...6.
->
Knowing change-
taking
->

so x & Yare indrand. Var

Than, Var(x+Y) =
Var(X) + Var(X)

variance of sum of many independent random variable:

The result that ind.var variance is equal to sum

ofvarianceholds fornotonlybutany rand

variat
an

ig*"="
#

var Var(X i)
c f
=
 standard deviation
To calculate standard daviation
of random variable:

of random var
we use $P(X) =
Varles (the sproot of variance (
Properties of std.dav:
SDCCX) =
23D(x)
5D(x+C)=5D(X)

Eg. VarKx) = 4, SDBX)=34=6

16 - Varlu)
Var(U) 480,
=

VarRx+33 16, 04 5DCX)=2

=
=

Condition Expectation Variance Stand. Dev

X Ex) v 1x) SD(X)

CX cECX) &VIX) CSD(X)

+C
* E(x) +9 vex) SD(X)

Examples:
Bernoullitial: A trial or an experiment whose
either
outcomes can be
classified as success or

Burnoulli
fail ure is trial
3:5 success, failures
which
let x be a rand. Var takes I as success

and bernoulli rand

0 as
failure. so X is Var.

Bernouli random variable: It is rand. Var. which

takes either 0 or bernoulli rand.
1. X is a Var. that
takevalue I with probability P means 0:1-P.
Probat dist
->

is:
L
x=xi)
O
* 1 -
P If
 ->
Expected
EX)=
value of berous and variable:
0x11-D) +PXP: P
->
Variance of bernoulli rand. Var
vCx P-p2=PC/-) =

Variance bernoul; distribution:

of
The when P=//2, when
largest variance occur

success&
failure equally likely.
are

word most uncertain bernoulli trials

In other
are those with
largest variance, resembles
tox of fair coin.

Independent distributed Bernoullitrials:

A collection
and
identically
of Bernoulli trials is defined as
ID Pernoulli trials.
are end. 9
A call
of
rand.Var Is IID if they
share a common prob.dist.
And trial I can each time
tossing
-

->

Nonind trial - of 3 ball without replacement

Choosing
pinomial random variable:
with resul
Suppose N ind trials are
performed
-

success' 'p'
'failure' with probablity 'l-p'
↑

as or
-

let X is total no
of success that vitrials,
occurs in
that X is binomial rand. Var with parameter n&P.
is
n = no of ind.trial (Heren fixed.)
P: Prob of success.
Binomial rand Var.:
X is binomial rand. Var with parameter n&P
that represent no
of successin ind. Demouli bials
when each trial is a success with prob.p.x take

within
value 0.1.
j the probableypini
Example A

ofdistribution:
binomial
Examples:
Expectation and variance
of binomial random variable
Binomial rand. Var is no
of success in 'n'indeptrial
when each trial is a success with prob. P.
Written as X =
Xt yet... +
Xn

From is sum of its

fact, exp. of sum
of rand var
E(x)
exp
Up
=

a, var. of sum
of ind.var is sum
of their variance:
Ver (x) =
npC1-P3

&| ⑨#

⑧| ↑
·

Hypergermetricdistribution
the
copulation
=>
must
bedded into one subsetoneis
and
without
Sampling replacement.
->

randomlyif we select is item without

replacement from set Nitem in which
of a

another
'm' items are a
type?'N-m are
type
N 2

- ~ -
~

2 N-m X 21-X

let x be no
of items of type 1, then PMt of
descrete rand Var X is called
hypergeometric dist

m N-m

P (x x)= 21 X
-

= X x =
0, 1, . . .
4
,
i

N ix 14sxkm, n-XXN-m
H

45. Ib X= 2
Best
↳ W, 2b

bits ↳O
W

YW, 3 b 30,2b X= (
W B
3
3w, 3b 3/b
F

L
3)
G
2 w, 3b X =
8
 Variance:

is
let x

objects are
a
hypergeometric
follow selected distribution in which
from N objects with in of object
in one
type aN-m objects
of being
second
type.
Vas(XS
M Nm
n
=
Nn
N-d
Poisson distribution: It probab of of events
give no

occuring in
fixed of
interval time or
space.

-Assumingtheseevents with known angrate,

happen he
↑

X denote no
of times an event occurs.

-we saye Poisson a usedas approxofbinol

is

age:
Probability mass
for of This distribution is
poisson:
is poison descrete
no of
per interval,
X events
ang
rand, var. Xv Poisson (x)
P(x=x) 2*x*, X: 0,1,2,
= ...

x
-
Rand Val X represent no. of interval per time interval
->e is mathematical constant 2. 718.
expectation of possonone variancetion of pain
on
continous random variable: It is the one that has
possible values that
form an interval
along real line.
It assume value in interval
can
any an

probablity density fr:

continous random variable
Every
associated
X has a curve

with
it.

Probability dist
of a continuous and
curve var. Is also
called probat density fr. denoted by FIM).
Area under PBF:
Consider ept.4's, where is less than 'b'
any
that assumes value that lies
The
prob a
blu
ab is
equal to area under curve
blew amb.
P(xE (a,b)) =P19xxxb) is area under curve byw a 9b.

=(u)
PlaxX1b) =

IEndn
Blue area b/w0&L.

a b

Area under entire curve

is 1.

Here open a close interval don't matter-

example: 10 d to no 500 t0

Cumulative dist.
for continous
In:
hand. Var X

F1al=P1x19) =

fLecxdx
Since, prob of cant rand Var X assume value
single
is always were so it will be same as above.

Expectation a variance:
Expected E=xf(xdx
value:
Variance:
(x-E)f(x)dx
Uniform distribution: A continous rand. Var has
a
uniform distribution, denoted XvU(a,b),
probab-density for is: 1/ (b -a) a <cb
Fus
fews S o otherwise

-
soi
 a = 2

b =
4

b-a = 2

hence,
t =
=
0.5

Standard uniform distribution:

A hand. Var has sad. uniform dist. With minor max
its prob.dens. Or
Jens
0 <RK1
if
M

given by o otherwise

Cumulative dist
of Uniform dist:for (a)

Eas I
for class

F
or (22 2b)
 rand.
Expectation & variance
I of uniform Var:

Expectation of (Xrvca,b);
E(X)= a+ B

*
erive:
ECX=/*xSCX, detoxdx tatr6=ba
Variance of X WU (a,b);
Var(x) =
(b-a) 2
12

CEXIT =
b-taltab (talba
D)
erives
Var(x) =
Gx2) -
t

Examplet

Answer
Non-uniform distribution: These are dust, which
not at all interval.
are same
point of time in a

distribution: To understand it do ex. of that

Trangular a

of time someone studied for 4 hr

exponential distribution: A
&
rand var whose pdf is

geven as
5cns<*e** xC,0
O otherwise W

is said to be
exp. distribution
L

with
parameter x.
 caf of exp. fn:
fas=P(x(9)
=

Jaxe dxa

=- ex o

= 1-ea

exponent and variance

of exp. distribution:
Nexp(X)
*
Ex) =
I

ECXY:
fe
VaRCE)= z
XR

Example: