Statistics
Statistics
Statistics
LECTURE MATERIAL
STATISTICS
TARUN KUMAR BANDYOPADHYAY, DEPARTMENT OF MATHEMATICS
In case you face difficulty in understanding the following material , you may e-mail
to me at
Chapter I
1
10
12
13
14
15
16
17
18
Tally marks:|
||
|||
||
||
||
||
||||
Frequency: 1
Marks:
21
22
24
25
26
27
28
29
31
33
34
Tally marks:|
||
||
||
||||
||||
|||
||
|||
||
||||
Frequency: 1
Marks:
35
36
38
42
47
Tally marks:
||
||
||
Frequency:
19
Tally Marks
0-5
6-10
11-15
16-20
21-25
26-30
31-35
36-40
41-45
46-50
///
/
////|
////|
////|
////|
////|
////|
/
/
//
////| /
/
////| ////| /
//
/
No.of
students
(frequency)
4
1
7
11
6
16
7
6
1
1
Cumulative
frequency(less
than)
4
5
12
23
29
45
52
58
59
60
Cumulative
frequency(greate
r than)
60
59
58
52
45
29
23
12
5
4
0+5
2 =2.5 which lies midway
between the lower and the upper limits is called the mid-value or central value of the
class. The less-than cumulative frequency (greater-than cumulative frequency
resp.) corresponding to a class is the total number of observations less than or equal to
the upper limit (greater than or equal to the lower limit) of the class. Following points
need be kept in mind while classifying given data:
Classes should be clearly defined and should not lead to any ambiguity
Classes should be exhaustive( each of given value should be included in one of
the classes)
Classes should be mutually exclusive and non-overlapping
Classes should be of equal width
Number of classes should neither be too large nor too small; preferably it
should lie between 5 and 15.
NOTE A variable, which can take any numerical value within certain range , is called a
continuous variable. Consider frequency distribution of the continuous variable of ages
in years of students in a college. We cannot arrange the data in age groups 16-20,21-25
etc. since there can be students having ages between 20 and 21 years. If the original
inclusive class intervals( of the form [a,b]) are ,say, 16-20,21-25,, we calculate the
adjustment
1
1
(lower
limit
of
succeeding
class-upper
limit
of
a
class)=
2
2 (21-20)=.5
Arithmetic Mean
Geometric Mean
Harmonic Mean
Median and Quartiles
4
Mode
Arithmetic Mean
If a variate X takes values x 1,,xn, then the A.M. of the set of observations x 1,,xn,
x 1 ++ x n
is defined by x =
. If the variate-values are not of equal importance, we
n
may attach to them weights w1,,wn as measures of their importance; the
corresponding weighted mean is defined by
x =
w1 x1 ++ wn x n
.
w 1+ + w n
f: 5
12
17
14
10
Computation of Mean
x
1
2
3
4
5
6
7
Total
f
5
9
12
17
14
10
6
73
fx
5
18
36
68
70
60
42
299
Thus
1
x =
N
f k x k= 299
73
1
0-10
10-20
20-30
30-40
40-50
50-60
No. of students: 12
18
27
20
17
Computation of Mean
Marks
No. of students(f)
0-10
10-20
20-30
30-40
40-50
50-60
Total
12
18
27
20
17
6
100
1
A.M.= x = N
Midpoint(x)
5
15
25
35
45
55
fx
60
270
675
700
765
330
2800
=28
f k x k = 2800
100
1
fx =a+h fu =a+ h u
f
f
xa
. Then
h
fx= af + hfu
. Thus
0-10
No. of students: 12
10-20
20-30
30-40
40-50
50-60
18
27
20
17
x300
Let us take the origin a=300 and scale h=50 so that u=
.
50
Properties of A.M.
Algebraic sum of deviations of a set of variate values from their arithmetic mean is
zero.
x ix
f i )=
f i x i x f i
=N x -N x
=0, where N=
fi
x 1
,,
xk
n
distribution of frequency N= i
of the combined
is given by:
1
x =
n x
N i=1 i i .
Example 1.4 The average salary of male employees in a firm was Rs. 5200 and
that of females was Rs. 4200. The mean salary of all the employees was Rs. 5000.
Find the percentage of male and female employees.
Let n1 and n2 denote respectively the number of male and female employees and
x 1
and
x 2
Geometic Mean
If n positive values x 1,,xn occur f1,,fn times respectively, then geometric mean(G.M.) G
of the set of observations is defined by G=
[ x1
f1
xn
1
fn N
, where N=
fi
1
Harmonic Mean
The harmonic mean H of n non-zero variate values x i with frequencies fi is given by H=
fi
f
xi
Median of a finite set of variate values is the value of the variate which divides it into two
equal parts. It is the value which exceeds and is exceeded by same number of
observations. Median is thus a positional average.
In case of ungrouped data, if the number of observations is odd then median is the middle
value after the values have been arranged in ascending or descending order of magnitude.
In case of even number of observations, there are two middle terms and median is taken
to be the arithmetic mean of the middle terms. Thus , median of the values
25,20,15,35,18, that is, of 15,18,20,25,35 is 20 and the median of 8,20,50,25,15,30,
that is, of 8,15,20,25,30,50 is (20+25)/2=22.5.
In case of discrete frequency distribution, median is obtained as follows:
fi
Find the cumulative frequency (less than type) just greater than N/2
The corresponding value of the variate is the median.
Example 1.5 Obtain the median for the following frequency distribution:
x:
f:
10
11
16
20
25
15
Calculation of Median
x
f
c.f.
1
8
8
2
10
18
3
11
29
4
16
45
5
20
65
6
25
90
7
15
105
8
9
114
9
6
120=N
N/2=60. The c.f. just greater than N/2 is 65 and the value of x corresponding to 65 is 5.
Thus , median is 5.
In case of grouped frequency distribution, median is obtained as follows:
Let us consider the grouped frequency distribution:
Class intervals
frequency
cumulative frequency
x1 - x2
f1
F1
x2 x3
f2
F2
.
8
xp xp+1
fp
Fp
..
xn xn+1
fn
Fn
where Fk=
fi
i=1
. Let the smallest c.f. greater then N/2 is F p. Then the median class is
xp xp+1. We assume that frequency of a class is uniformly distributed over the class
interval. Let the c.f. for the class just above the median class be c . Thus (N/2-c) is the
frequency of the interval between the median and the lower limit of the median class . the
length of the interval corresponding to the frequency
(N/2-c) is
N
c
2
f
I, where f is
frequency of the median class, I is the length of the class interval of the median class .
N
c
2
Hence the median is L0+
f
Properties of Median
Quartiles are thsose variate values which divide the total frequency into four equal parts;
deciles and percentiles divide into ten and hundred equal parts respectively. Suppose
the values of the variate have been arranged in ascending order of magnitude, then the
value of the quartile having the position between the lower extreme and the median , is
the first quartile Q1 and that between the median and the upper extreme is the third
quartile Q3. The median is the second quartile Q 2, is the fifth decile D5 and the fiftieth
percentile P50. For a grouped frequency distribution, the quartiles, deciles and percentiles
are given by
C
4
Qi=l+
h, i=1,2,3
f
jN
C
Dj= l+ 10
h, j=1,,9
f
kN
C
Pk= l+ 100
h, k=1,,99
f
where l is the lower limit of the class in which the particular quartile/decile/percentile lies,
f is the frequency of the class , h is the width of this class, C is the cumulative frequency
upto and including the class preceding the class in which the particular
quartile/decile/percentile lies and N is the total frequency.
Example 1.6 Calculate the three quartiles for the following frequency distribution of the
number of marks obtained by 49 students in a class:
Marks
No. of students
Marks
No. of students
5-10
25-30
10-15
30-35
15-20
15
35-40
20.25
10
40-45
Frequency
5-10
10-15
11
15-20
15
26
20-25
10
36
25-30
41
30-35
45
35-40
47
40-45
49=N
The cumulative frequency immediately greater than N/4=49/4 is 26; hence to find Q 1,
10
49
11
L=15, h=15-10=5, C=11, f=15. Thus Q1= 15+ 4
15
5= 15.47 marks.
49
11
For median, N/2= 24.5 . Thus the median class is 15-20. Median = 15+ 2
15
5=19.5
marks.
To find Q3, we have 3N/4=147/4 . Hence Q 3 lies in the class 25-30. L=25, C=36,f=5,h=5.
147
36
4
Hence Q3=25+
5 =25.75.
5
Example 1.7 In a factory employing 3000 persons, in a day 5% work less than 3 hours,
580 work from 3.01 to 4.50 hours, 30% work from 4.51 to 6.00 hours, 500 work from
6.01 to 7.50 hours, 20% work from 7.51 to 9.00 hours and the rest work 9.01 or more
hours. What is the median hours of work?
Calculation for Median Wages
Work Hours
Less than 3
3.01-4.50
4.51-6.00
6.01-7.50
7.51-9.00
9.01 and above
No. of employees(f)
5/100 x 3500=150
580
30/100 x 3000=900
500
20/100 x 3000=600
3000-2730=270
Class Boundaries
Below 3.005
3.005-4.505
4.505-6.005
6.005-7.505
7.505-9.005
9.005 and above
N=3000. The c.f. just greater than N/2=1500 is 1630.the corresponding class 4.51-6.00,
whose class boundaries are 4.505-6.005, is the median class. Hence median=l+
h N
c
2 2
1.5
(1500730)
=4.505+ 900
=4.505+1.283=5.79(approx.).
Frequency
Variable
10-20
12
50-60
20-30
30
60-70
25
30-40
70-80
11
Frequency
18
40-50
65
Total(N)
229
Given that the median value is 46, determine the missing frequencies.
Let the frequency of the class 30-40 be f 1 and that of 50-60 be f 2. Then
f1+f2=229 -(12+30+65+25+18)= 79.
Since median is given to be 46, 40-50 is the median class. Using formula for
median , we get
114.5(12+30+f 1)
46=40+
x 10. Hence f1= 33.5= 34(approx.). Hence
65
f2=45.
Mode
Let us consider the following statements: The average height of an Indian is 56;
the average size of shoes sold in a shop is 7; an average student in a hostel
spends Rs. 750 per month. In all the above statements,the average referred to is
mode. Mode is the value of the variate which occurs most frequently in a set of
observations and around which the other members of the set cluster densely. In
other words, mode is the value of the variable which is predominant in the given
set of values. In case of discrete frequency distribution, mode is the value of the
variable corresponding to maximum frequency. In the following distribution:
x: 1 2
f: 4 9
16
25
22
15
10
11
12
Frequency: 3
15
23
35
40
32
28
20
45
14
The distribution is not regular since the frequencies are increasing steadily
upto 40 and then decrease but the frequency 45 after 20 does not seem to
be consistent with the distribution. We cannot say that since the maximum
12
Size
(x)
1
(i)
3
(ii)
Frequency
(iii)
(iv)
(v)
(vi)
11
26
23
3
15
46
38
23
35
58
98
73
75
6
40
107
72
32
28
20
60
80
100
48
65
10
93
45
79
59
11
14
12
65
20
The frequencies in column (i) are the original frequencies. Column (ii) is obtained
by combining the frequencies two by two.If we leave the first frequency and
combine the remaining frequencies two by two, we get column (iii).Combining the
frequencies two by two after leaving the first two frequencies results in a
repetition of column (ii). Hence, we proceed to combine the frequencies three by
three , thus getting column (iv). The combination of frequencies three by three
after leaving the first frequency results in column (v) and after leaving the first
two frequencies results in column (vi).
The maximum frequency in each column is given in red type. To find mode we
form the following table:
13
Analysis Table
Column No. Maximum Frequency(1)
Where l is the lower limit of the modal class(class having maximum frequency), f m
is the maximum frequency, f 1 and f2 are the frequencies of the classes preceding
and following modal class.
Example 1.10 The median and mode of the following wage distribution are known
to be Rs. 3350 and Rs. 3400 respectively.Find the values of f3,f4,f5:
Wages (in Rs.) No. of employees
0-1000
4000-5000
f5
1000-2000
16
5000-6000
2000-3000
f3
6000-7000
3000-4000
f4
Total
230
(in Rs.)
frequency(f)
0-1000
1000-2000
16
20
2000-3000
f3
20+f3
3000-4000
f4
20+f3+f4
14
4000-5000
f5
20+f3+f4+f5
5000-6000
26+f3+f4+f5
6000-7000
30+f3+f4+f5=N
1000
f 4 [115-(20+f3)]. Thus f3=95-0.35f4.
Mode being 3400, modal class is 3000-4000. Using formula for mode,
3400=3000+
1000 (f 3 f 4)
2 f 4f 3f 5
; hence
f 4 +0.35 f 495
34003000
=
1000
2 f 4(200f 4) . Thus f4=100.
Chapter II
Frequency Distributions and their Comparison:
Measures of Dispersion
A measure of central tendency alone is not enough to have a clear idea about the
data unless all observations are almost the same. Moreover two sets of
observations may have the same central tendencies whereas variability of data
within the sets may vary widely . Consider
Set A:
30
30
30
30
30
Set B:
28
29
30
31
32
Set C:
30
37
75
15
All the three sets have same mean and mode; but the amount of variation differs
widely amongst the sets.
Characteristics of an ideal measure of dispersion
It
It
It
It
It
should
should
should
should
should
be
be
be
be
be
rigidly defined
easily understandable and easy to calculate
based on all observations
amenable to further mathematical treatment
least affected by fluctuation of sampling
1
2
(Q3-Q1), where Q1 and Q3 are the first and the third quartiles of the frequency
distribution. Quartile deviation is definitely a better measure than the range as it
makes use of 50% of data. But since ignores the other 50% of data, it cannot be
regarded as a reliable measure.
Mean Deviation If xi|fi ,
1
f |x A| ,
N i=1 i i
fi
i=1
=N.
0-10 10-20
No. of students: 6
20-30
30-40
40-50
15
16
50-60
6
60-70
3
Marks
0-10
10-20
20-30
30-40
40-50
50-60
60-70
Total
(1)
Midvalue
5
15
25
35
45
55
65
6
5
8
15
7
6
3
d=(x35)/10
fd
-3
-2
-1
0
1
2
3
-18
-10
-8
0
7
12
9
-8
|xx|
|xx|
28.4
18.4
8.4
1.6
11.6
21.6
31.6
170.4
92.0
67.2
24.0
81.2
129.6
94.8
659.2
c.f.
(less)
6
11
19
34
41
47
50
The c.f.(less than) just greater than 12.75 is 19. Hence the corresponding class
20-30 contains Q1.
Q1=20+
10
8 (12.75-11)=22.19
The c.f.(less than) just greater than 37.25 is 41. Hence the corresponding class
40-50 contains Q3.
10
Q3=40+ 7 (37.25-34)=44.64.
Hence Q.D. =
(2)
1
1
2 (Q3-Q1)= 2 (44.64-22.19)=11.23.
x = A+
h fd
10 X (8)
=35+
50
N
1
659.2
f |x x |
mean) = N
= 50 =13.184.
Standard Deviation, Variance
17
1
f i (x i x )2 , where
i=1,,n , S.D.
is defined by:
xi|fi ,
1
1
1
f i (x i A)2
f i (x i x + x A)2
f {( x x )2
s= N
= N
= N i i
+( x -A)2+2(
2
x -A)(x - x )}
i
1
1
1
f (x x )2
f
f ( x x )2
= N i i
+( x -A)2 N i +2( x -A) N i i
=
+( x -A)2,
where d= x -A.
Thus s2 is least when d=0.that is, when
Note
(1)
x 2i + x 22 x i x
1
fi
f (x x )2
2=
=
)
N i i
1
1
1
f x2
x 2
N i i +
N
1
1
1
1
2
2
2
f i xi
f i x 2i
f
x
x
x
f i xi
i
i
= N
+
-2
= N
N
N
- x
and
x =
. This expression is
.
x 1
and
x 2
x 2
fi
and
21 +d 21
n1
22 +d 22
x
2=
)+ n2 )], where d1= 1 - x , d2=
1
n1 + n2
n1 x1 +n2 x2
n1 +n2
18
Example 2.2 For a group of 200 candidates , the mean and S.D. of scores were
found to be 40 and 15 respectively. Later on it was discovered that the scores 43
and 35 were misread as 34 and 53 respectively. Find the corrected mean and S.D.
corresponding to the corrected figures.
Let x be the given variable. Given n=200,
1
x
x =
200 i
gives
1
x2
= n i - x
Corrected
xi
=8000.
x 2i
gives
x =40 and
=200(225+1600)=365000.
7991
=8000-34-53+43+35=7991, corrected mean = 200
xi
=39.995
Corrected
x 2i
Corrected
= 365000-342-532+432+352=364109
364109
=
(39.995)2=224.14.
200
Thus
corrected
224.14 =14.97.
Example 2.3
The first of two samples has 100 items with mean 15 and s.d. 3.if
the whole group has 250 items with mean 15.6 and s.d. 13.44 . find s.d. of the
second group.
Here n1=100,
x =
x 2
x 1
n1 x1 +n2 x2
n1 +n2
=15,
gives
=3, n=n1+n2=250,
x 2
- x =16-15.6=0.4
2
From
1 +d 1
2
2
n1
2 +d 2
2=
)+ n2 )],
1
n1 + n2
Moments
19
=4.
x =15.6,
x 1
13.44
- x =15-15.6=-0.6, d2=
given by
, is
, is
1
f (x A)r
= N i i
.
, denoted by
1
f (x x )r
= N i i
.
0
In particular,
1
1
0
f i ( x i x )
f
= N
= N i =1,
1
f ( x x )2
=
N i i
1
1
f i ( x i x )
= N
=0,
1
1
1
f (x x )r
f (x A + A x )r
f (d + Ax )r
= N i i
= N i i
= N i i
, where di=xi-A.
r
1
f d
x =A+
N i i =A+ 1 . Hence
di C 1
1
r
r r2
2
f
r
f
(d
)
i
= N i i 1 =
+ C2 d i ( 1 )
1
r
-+(-1)r ( 1) )
r
2
r
- C1 r1 1 + C2 r2 ( 1 ) -+(-1)r (1) .
-3
3
+2 1 ,
4 4 3 1
2
4
+6 2 1 -3 1 .
20
2
- 1 ,
Let
u=
x A
.
h
Then
x = A+h u .
Thus
x- x =h(u- u ).
Thus
r (x )
f i {h ( uiu ) } =
1
r
f i (x i x ) =
(u)
N
hr r
.
1
f:
10
1
f ( x x )n
N 1 i i
23
1= 3
2
= 0. Thus
symmetry.
Also for a symmetrical distribution, the mean, median and mode coincide. Further,
in the case of such distribution, median lies halfway between the two quartiles.
Skewness means lack in symmetry. It indicates whether the frequency curve is
inclined more to one side than the other, that is , whether the frequency curve has
a longer tail on one side. Skewness is positive if the curve is more elongated to
the right side, that is, if the mean of the distribution is greater than the mode; in
the reverse case, it is negative. Skewness gives an idea about the direction in
which also the extent to which the distribution is distorted from the symmetrical
distribution.
For distribution of moderate skewness, an empirical relation holds: mean-mode=
3(mean-median).
21
Q3 +Q12Q2
.
Q3Q 1
Example 2.4 Find out a coefficient of dispersion based on quartile deviation and
a measure of skewness from the following table giving wages of 230 persons:
Wages(in Rs)
c.f.
Wages(in Rs)
c.f.
140-160
12
12
220-240
50
157
160-180
18
30
240-260
45
202
180-200
35
65
260-280
20
222
200-220
42
107
280-300
230
Here N/2=115 and the 115 person has a wage in the class 220-240. Hence
115107
median=Q2= 220+
x 20 = Rs. 223.20. Similarly, Q1=180+
50
57.530
172.5157
x 20=Rs. 195.71, Q3=240+
x 20=Rs. 246.88. It can
35
45
be shown that mean=Rs. 220.87, S.D.= Rs. 34.52.
Coefficient of dispersion based on quartile deviation=
=0.1156.
Measure of skewness =
meanmode
220.87233.00
=
= - 0.3514
S. D.
34.52
Q3 +Q12Q 2
Second measure of skewness=
=- 0.07446.
Q3Q 1
Pearsons
and
Q 3 Q1
51.17
Q 3+ Q1 = 442.59
- Coeficients
22
1= 33
2
1= 1
2=
4
22
-3.
2
distribution is symmetrical and whether it is relatively more or less flat than the
1
normal curve.
gives a measure of departure from symmetry. Kurtosis
measures whether the given frequency curve is relatively more or less flat topped
compared to the normal curve (to be studied later). For a normal distribution ,
2
2
=3. Curves with values of
less than 3 are called Platykurtic whereas
those with values of
Chapter III
Theory of Probability
Basic Terminology
Random Experiment : If in each trial(repetition) of an experiment conducted
under identical conditions, the outcome is not unique , but may be any one of
possible outcomes, then such an experiment is called a random experiment.
Examples of random experiments are: tossing a coin, throwing a die, selecting a
card from a pack of playing cards etc. in all these cases, there are a number of
possible outcomes which can occur but there is an uncertainty as to which of them
will actually occur.
A piece of Information: A pack of cards consists of four suits called Spades,
Hearts and Clubs. Each suit consists of 13 cards, of which nine cards are
numbered from 2 to 10, an ace, a king, a queen and a jack(or knave).Spades and
clubs are black-faced cards while hearts and diamonds are red-faced cards.
Outcome: result of a random experiment is called an outcome.
Sample Space, Events: The collection S of all possible outcomes of a random
experimepernt is called sample space of the random experiment; any subset of S
is an event; a singleton subset of S is an elementary(simple) event. For example,
in an experiment which consists of throwing a six-faced die, possible outcomes are
1,2,3,4,5,6. Thus sample space of this experiment is {1,2,3,4,5,6}, {1} is an
23
Equally Likely Events: Events are equally likely if there is no reason to expect
any one of them compared to others. In the trial of drawing a card from a wellshuffled pack of cards, any of the 52 cards may appear, so that the 52 elementary
events are equally likely.
Exclusive Events: Events are exclusive if the occurrence of any one of them
precludes the occurrence of all others. On the contrary, events are compatible if it
is possible for them to happen simultaneously. For instance, in the rolling of two
dice, the cases of the face marked 5 appearing on both dice are compatible.
Favourable Events: The trials which entail the happening of an event are
favourable to the event. For example, in the tossing of a dice, the number of
favourable events to the appearance of a multiple of 3 are two viz. getting 3 and
6.
Classical (a priori) definition of probability
If a random trial results in n exhaustive, mutually exclusive and equally likely
outcomes, out of which m are favourable to the occurrence of an event E, then the
m
probability of occurrence of E , denoted by P(E), is given by P(E)= n .
It is clear from definition that 0 p 1. Since the number of cases in which
event A will not happen is n-m, the probability q that the event A will not happen
nm
m
is given by P ( A ) =
=1- n =1-P(A).
n
An event A is certain to happen if all the trials are favourable to it and then the
probability of its happening is unity; for an event which is certain not to happen,
the probability is zero.
Example 3.1 Find the chance that if a card is drawn at random from an ordinary
pack, it is one of the court cards.
24
Court cards are kings, queens, jacks and their number in a pack of 52 cards
is 12, so that the number of favourable events is 12. Hence the probability is
12/52=3/13.
Example 3.2
53 Sundays?
What is the chance that a leap year selected at random will contain
equally likely, exhaustive and mutually likely cases. A red ball can be drawn
in 2 ways, a blue in 3 and a black in 4 ways, so that three differently
coloured balls can be drawn in 2 x 3x 4=24 ways. Hence the probability is
24/84=2/7.
(2)two blue balls can be drawn in
C2
ways so that the two blue balls and a different coloured ball can be drawn in
3
6x C2 =18 ways. Two black balls and a different coloured ball can be drawn
4
in 5x C2 =30 ways. Similarly the number of ways in which two red balls and
2
a different coloured ball can be drawn in 7x C2 =7 ways. Thus the number
of ways two balls of same colour and a ball of different colour can be drawn
is 18+30+7=55. Thus required probability is 55/84.
(3)Three blue balls can be drawn in 1 way and 3 black balls in
ways so that the corresponding probability is 5/84.
Limitation of Classical Definition
25
C3
or 4
If the various outcomes of the random experiment are not equally likely
If the number of exhaustive outcomes of the random experiment is infinite
or unknown
Von Misess statistical (or empirical) definition of probability
(Axiom of non-negativity)P(A) 0
events, then P ( n A i )= P (A i ) .
i=1
A B=( A B)+(A B ).
(1)Only A occurs, (2) Both A,B but not C , occur, (3) All three events occur, (4) At
least one occurs, (5) At least two occur, (6) one and no more occurs, (7) two and
no more occur, (8) none occurs.
26
(1) A B C
B C )
A B C
C ( A
B C) ),
A B
A B C )
( A B C ),
( A B C )
A B C
A B C
A B C
(7)
)
)
),
(8)
(6)
A B C .
Some Theorems on Probability
Theorem3.1 Probability of impossible event is zero: P( )=0.
S=S
and S,
Theorem3.2 P( A =1-P(A).
1=P(S)=P(A A )=P(A)+P( A
(since A A = ).
A B ( A )
and
A B ( A )
B)
(A A )=( A B ) ( A
; hence by axiom of additivity, P(
A B =P(B)P( A B) .
27
A B=A ( A B ) and A,
A B
P( A B = P(A)+P(B)-P( A B ).
Generalising , for three events A,B,C, we have
P( A B C =P( A)+ P(B)+ P( C)P( A B)P(B C)P(C A )+ P (A B C) .
If p1=P(A), p2=P(B), p3=P( A B ), express the following in terms
(1)
A B )=P( A B
)=1-P( A B)=1-p3.
B)=p2-p3.(4)
P( A B )=P( A )+P(B)-
(3)
P( A B)=P(B)-
P( A B)=1-p1+p2-(
P(
P( A
p2-p3)=1-
p1+p3.
Example 3.6 It two dice are thrown, what is the probability that the sum is (1)
greater than 8, (2)neither 7 nor 11?
(1)If X denotes the sum on the two dice, then we want P(X>8). The required
event can happen in the following mutually exclusive cases: X=9,
X=10,X=11,X=12.
Hence
by
addition
theorem
on
probability,
P(X>8)=P(X=9)+P(X=10)+P(X=11)+P(X=12). In a throw of two dice, the
sample space contains 6 2=36 points. The number of favourable cases can be
enumerated as:
X=9: (3,6),(6,3),(4,5),(5,4)
X=10: (4,6),(6,4),(5,5)
X=11: (5,6),(6,5)
X=12:(6,6).
4 3 2 1
5
+ + +
Thus P(X>8)= 36 36 36 36 = 18 .
(2)Let A denote the event of getting the sum of 7 and B denote the event of
getting the sum of 11 with a pair of dice.
X=7: (1,6),(6,1),(2,5),(5,2),(3,4),(4,3)
X=11: (5,6),(6,5)
28
Required probability= P( A B )=1-P(A B)=1-[P(A)+P(B)] (since A and
1 1
7
18 5
3
5
+
Required probability is P(A B)=P(A)+P(B)-P(A B)= 36 36 36 = 9 .
Example 3.8 An integer is chosen at random from first two hundred natural
numbers. What is the probability that the integer is divisible by 6 or 8?
Sample space of the random experiment is {1,2,,200}. The event A
integer chosen is divisible by 6 is given by {6,12,,198}; the event B
integer chosen is divisible by 8 is given by {8,16,,200}. LCM of 6 and 8 is
24. Hence a number is divisible by 6 and 8 iff it is divisible by 24. Thus A
B={24,48,,192}. Hence required probability is P(A B)=P(A)+P(B)-P(A
33 25
8
1
+
B)=
200 200 200 = 4 .
Example 3.9 The probability that a student passes Physics test is 2/3 and the
probability that he passes both Physics test and English test is 14/45.The
probability that he passes at least one test is 4/5. What is the probability that he
passes English test?
Let A be the event that the student passes the Physics test and B be the
2
14
29
4
B)=
B)=P(A)+P(B)-P(A B), we get
5 . We want P(B). From P(A
4
P(B)= 9 .
Example 3.10 An investment consultant predicts that the odds against the price
of a certain stock will go up during the next week are 2:1 and the odds in favour of
the price remaining the same are 1:3.What is the probability that the price of the
stock will go down during the next week?
Let A denote the event stock price will go up and B be the event stock
1
1
price will remain same. Then P(A)= 2+1 , P(B)= 1+3 . Thus P(A
1 1 7
+ =
B)=P(A)+P(B)= 3 4 12 . Hence the probability that the stock price will go
7
5
down is given by P( A B )=1- P(A B)=1- 12 = 12 .
Example 3.11 An MBA applies for a job in two firms X and Y. The probability of
his being selected in firm X is 0.7 and being rejected at Y is 0.5. The probability of
at least one of his applications being rejected is 0.6. What is the probability that
he will be selected in one of his firms?
Let A and B denote the events that the person is selected in firms X and Y
0.5=0.5 and 0.6= P( A B )= P ( A ) + P ( B )
P( A B ). The probability
that the person will be selected in one of the twofirms X or Y is given by: P(A
) +
B)=1- P( A B )=1-[ P ( A
P ( B )
P( A B ]=1-(0.3+0.50.6)=0.8.
Example 3.12 Three newspapers A,B and C are published in a certain city. It is
estimated from a survey that 20% read A, 16% read B, 14% read C, 8% read
both A and B, 5% read both A and C, 4% read both B and C, 2% read all three.
Find what percentage read at least one of the papers?
30
Let E,F,G denote the events that a person reads newspapers A,B and C
20
16
14
respectively. Then we are given: P(E)= 100 , P(F)= 100 , P(G)= 100 , P(E
8
5
4
E F G
F)=
G)=
F)=
)=
100 , P(E
100 , P(G
100 , P
2
100 .
The required probability that a person reads at least one of the newspapers
is given by
E F G
P
)=P(E)+P(F)+P(G)- P(E F)- P(E G)- P(G F)+ P
35
E F G
)= 100 =0.35.
Example 3.13 A card is drawn from a pack of 52 cards. Find the probability of
getting a king or a heart or a red card.
Let A,B and C denote the event card drawn is a king, card drawn is a heart
and card drawn is a red card respectively. Then A,B,C are not mutually
exclusive.
A B: card drawn is king of hearts ; n(A B)=1
B C =B( since BC): card drawn is a heart ;n(B C )=13
A C : card drawn is a red king; n(A C )=2
A BC = A B : card drawn is the king of hearts; n(A BC )=1.
4
13
26
1
BC
)= 52 , P(A
)= 52 . Thus required probability is P(
52 , P(A
A B C )=P(A)+P(B)+P(C)-
P(A B)-
P(B C )-
P(A C )+
P(A
7
BC )=
13 .
Conditional Probability
In many situations we have the information about the occurance of an event A and
are required to find out the probability of the occurrence of another event B. This
is denoted by P(B/A). For example, if we know that a card drawn from a pack is
black, we may need to calculate the probability that it is the ace of spade.
31
Let us take the problem of throwing a fair die twice. Suppose same number of
spots do not appear in both the throws and we are required to find the probability
that the sum of number of spots in the two throws is six.
A patient comes to a doctor with his family history that his elders suffered from
high blood pressure. He wants to know the probability of the event that he will
also suffer from high blood pressure.
Definition Let A and B be two events. The conditional probability of event B
P( A B)
supposing event A has occurred, is defined by P(B/A)=
, if P(A)>0.
P( A)
Note Let B1,,Bk be mutually exclusive events. The conditional probability of
1
k Bi
1 k Bi
1
given that event A has occurred is given by P
=
P[ ( k B i ) A ]
( A )=
P (A )
k
P (B i A)
1
P( A)
Example 3.14 If a card drawn from a pack is black, represented by event A, find
the probability of the event B that the card drawn is ace of spade .
Number of black cards in a pack of 52 cards is 26. P(A)=26/52=1/2. Out of
26 black cards, only one is ace of spade. The event A B contains only one
point; thus
1
P( A B)
1
52
1
P ( A B)=
.
Hence
P(B/A)=
=
=
52
1
26 .
P( A)
2
4
36
4
(1,5),(2,4),(4,2),(5,1). Thus P(A B)=4/36. Thus P(B/A)= 30 = 30
36
=0.133.
Example 3.16 10% of patients feel they suffer and are really suffering from TB,
30% feel they suffer but actually do not suffer, 25% do not feel they are suffering
but are suffering and remaining 35% neither feel nor suffering from TB. Find the
probility of events E1,E2,E3,E4, where E1: person who suffers from TB and feels he
suffering from TB, E2: person has TB and does not feel, E 3:person feels he has TB
and does not suffer from TB, E4: person feels and has TB.
Let us define events: A: person feels he has TB, B: person suffers from TB.
P(A B)=0.1, P(A B =.3, P( A B)=.25, P( A B )=.35. Thus P(A)=
P(A B)+
P(A B =0.1+0.3=0.4,
P(B)=
P(A B)+
P( A
P(A B )+ P( A B )=0.3+0.35=0.65. Hence
P( A B) 0.1
=
=0.25
P(E1)=P(B/A)=
,
P( A)
0.4
P(B A)
A
P(E2)=P(B/
)=
)
P( A
Example 3.17 There are two lots of manufactured item. Let one contain 40 pieces
whereas lot two contains 50 pieces.it is known that former lot contains 25%
defective pieces and the later one 10%. We flip a coin and select a piece from a lot
one if it turns with head up; otherwise we select a piece from lot 2. Find the
probability that a selected piece will be defective.
Let A,B stand for the event piece is selected from lot 1 and piece is
selected from lot 2. Since the probability of turning head up=1/2, we have
33
A )=1/20. Hence the probability that the selected item is defective is P(A
B)+ P( A B)=0.175.
Independent Events
If we draw two cards from a pack of cards in succession, then the results of the
two draws are independent if the cards are drawn with replacement and are not
independent if the cards are drawn without replacement.
Definition An event A is independent of another event B iff P(A/B)=P(A). This
definition is meaningful when P(A/B) is defined, that is, when P(B) 0.
Theorem3.4 If two events A and B are such that P(A) 0, P(B) 0 and A is
independent of B, then B is independent of A.
P( A B)
=P( A)
P(A/B)=P(A)
P( B)
P( A B) =P(A)P(B) P(B/A)=
P( A B)
=
P( A)
P ( A ) P(B)
=P(B).
P( A)
If A,B are events with positive probilities, then A and B are
independent iff P( A B) =P(A)P(B).
Theorem3.5
B , (2)
and B ,
P( A B) =P(A)P(B). P(A B
)=P(A)-
P( A B) =P(A)-P(A)P(B)=P(A)P( B ).
P( A B )=P( A B )=1-P(A
A B
B)=1-[P(A)+P(B)- P )]=1-P(A)-P(B)+P(A)P(B)=[1-P(A)][1-P(B)]=P(
A )P( B ).
Example 3.18 If
34
AB
A B
A=( A B )= )= A B
A B P(A) P( B .
Example 3.19 Let A and B be two events such that P(A)=3/4, P(B)=5/8. Show
3
3
5
P ( A B)
that (a) P(A B)
,
(b)
4
8
8 .
3
(a) AA B 4 P( A) P( A B) .
5
A B BP( A B P(B)=
8 .
(b)
Also,
A
P ( A B )=P ( A ) + P ( B )P B)
A B
3
P ( A B)
3 5
1 + 1 P ) 8
4 8
BAYES THEOREM
Theorem3.7 If E1,E2,,En are mutually disjoint events with P(Ei) 0(i=1,,n),
then for any arbitrary event A which is a subset of
P ( Ei ) P(
P(Ei/A)=
A
)
Ei
A
P ( E i ) P( E )
1
i
1 n Ei
A
)
Ei
P( A)
P ( Ei ) P (
=
Example 3.20 Suppose that a product is produced in three factories X,Y, and Z. It
is known that factory X produces thrice as many items as factory Y and that factory
Y and Z produces same number of items. 3% of the items produced by each of the
factories X and Z are defective and 5% of those manufactured in Y are defective.
All the items in the three factories are stocked and an item of the product is
selected at random. (1) What is the probability that the item selected is defective?
(2) if an item selected at random is found to be defective, what is the probability
that it was produced in factory X,Y,Z respectively?
Let the number of items produced by factories X,Y, and Z be 3n,n,n
respectively. Let E1,E2,E3 be the events that the items are produced by
factory X,Y and Z respectively and let A be the event that the item being
35
3n
defective. Then P(E1)= 3 n+ n+n =0.6, P(E2)=0.2, P(E3)=0.2. Also, P(A/E1)=
P(A/E3)=0.03, P(A/E2)=0.05(given).
(1)The probability that an item selected at random from the stock is
3
defective
is
given
by
A
P ( A Ei ) = P ( Ei ) P ( ) =.6
P(A)=
E
1
1
i
03+.2x.05+.2x.03=.034.
(2)By Bayes rule, the required probabilities are given by :
A
A
P ( E1 ) P( )
P ( E2 ) P( )
9
E1
E2
.6 X .03 =
.2 X .05 =
P(E1/A)=
P(E2/A)=
=
17 ,
=
P( A)
.034
P( A)
.034
A
P ( E3 ) P( )
5
E3
3 .
17 , P(E3/A)=
=
P( A)
17
Example 3.21
In 2002 there will be three candidates for the position of the
principal Mr. x, Mr. y and Mr.zwhose chances of getting the appointment are in
the ratio 4:2:3 respectively. The probability that Mr. x if selected would introduce
co-education in the college is 0.3. The corresponding probabilities for Mr. y and
Mr.z are 0.5 and 0.8. (1) What is the probability that there will be co-education in
2003? (2) if there is co-education in 2003, what is the probability that Mr. z is the
principal?
Let us define the events
A: introduction of co-education, E1: Mr. x is selected as principal
E2: Mr. y is selected as principal, E3: Mr. z is selected as principal.
P(E1)=4/9, P(E2)=2/9, P(E3)=3/9, P(A/E1)=3/10, P(A/E2)=5/10, P(A/E3)=8/10.
(1)The required probability that there will be coeducation in the college in
2003 is
P(A)=P[(A E1) ( A E 2)( A E3 ) ]= P(A E1)+ P(A E2)+ P(A
E3 )
4 3 2 5 3 8
23
=P(E1)P(A/E1)+ P(E2)P(A/E2)+ P(E3)P(A/E3)= 9 . 10 + 9 . 10 + 9 . 10 = 45 .
(2)The required probability is given by Bayes rule:
36
P(E3/A)=
P ( E3 ) P(
A
)
E3
P( A)
3 8
X
9 10
12
=
=
46
23 .
90
Chapter IV
Random Variable and Distribution Function
In many random experiments, we are interested not in knowing which of the
outcomes has occurred but in the numbers associated with them. For example,
when n coins are tossed, one may be interested in knowing the number of heads
obtained. When a pair of dice are tossed, one may seek information about the sum
of points. Thus, we associate a real number with each outcome of a random
experiment. In other words, we are considering a function whose domain is the set
of all possible outcomes and whose range is a subset of the set of reals.
Definition Let S be the sample space associated with a given random experiment.
,
A real-valued function X: S ) is called a one-dimensional random
variable(r.v.).
Notation If x is a real number, the set of all w S such that X(w)=x is denoted
by X=x. Thus P(X=x)=P{w: X(w)=x}. Similarly P(X a)= P{w S : X (w) },
P(a<X b)=P{w:X(w) (a,b]}
Example 4.1 Consider the random experiment of tossing a coin. Then S={w 1,w2},
w =H,w =T. Define X:{w ,w } {0,1} by X(w )=1, X(w )=0. X is a r.v.
1
P(X=1)=P{(i,j):X(i,j)=1}=P{(1,1)}=1/36, P(X=2)=P{(1,2),(2,2),(2,1)}=3/36.
Distribution Function
Definition Let X be a random variable(r.v.). The function F : (-,) [0,1]
defined by F(x)=P{t: X(t)x} is the distribution function (d.f.) of the r.v. X.
Note: To emphasize the r.v. X, we sometimes denote F(x) by FX(x).
Properties of Distribution Function
(1)
(2)
p ( x i )=1.
1
38
Example 4.4 Let S={H,T} be the sample space corresponding to the random
experiment of tossing of a fair coin. Let X be the r.v. defined by X(H)=1, X(T)=0.
X has only two distinct values, namely, 0 and 1. The corresponding p.m.f. is given
by: p(1)=P(X=1)=P(H)=1/2, p(0)=1/2.
Example 4.5 A r.v. X has the following p.m.f.:
xi:
pi:
2k
2k
3k
k2
2k2
7k2+k
Example 4.6
P{
P( X >1)
P ( X >1)
39
2
P( X=2)
15
1
=
=
1 7
1P( X=1)
1
15
Random Variables
Y
Z
X+Y
XY
HHH
HHT
HTH
HTT
THH
THT
TTH
TTT
3
2
2
1
2
1
1
0
1
1
0
0
1
0
0
0
3
2
0
0
2
0
0
0
4
3
2
1
3
1
1
0
3
2
0
0
2
0
0
0
(1)Obviously X is a r.v. which can take the values 0,1,2, and 3. p(3)=P(HHH)=
3
1
1
=
p(2)=P(HHT HTH THH )=P(HHT)=P(HTH)
8 ,
2
()
of
U=X+Y:p(0)=1/8,
40
The probability for a variate value to lie in the interval [a,b] is P(a X b)=
b
f X ( x ) dx
a
f(x) 0,
f ( x ) dx
=1
(since
f ( x ) dx
gives
total
probability),
P(X=c)=
f X ( x ) dx
c
A.M.
x =
xf ( x ) dx ,
a
(central)=
( xx )r f ( x ) dx
a
r (about
x=A)=
( x A )r f ( x ) dx
a
Median is the point which divides the total area into two equal parts: if M be the
M
median, then
f ( x ) dx= f ( x ) dx=1/2.
a
Mode is the value of x for whixh f(x) is maximum. Mode is the solution of f /(x)=0
and f//(x)<0, provided it lies in [a,b].
Example 4.8
41
(1)
f ( x ) dx=1
0
f ( x ) dx= 6 f ( x ) dx
b
P(X<b)=P(X>b)
6
0
Example 4.9
(1)p=P(X 150)=
150
f ( x ) dx=
100
100
100
1
dx= .
2
3
x
By
compound
probability
theorem, the probability that all three of original tubes will have to be
replaced during the first 150 hours =p3=1/27.
(2)The probability that a tube is not replaced during the first 150 hours of
operation is P(X>150)=1-P(X 150)=1-p=2/3. By compound probability
theorem, the probability that none of the three tubes will have to be replaced
during the first 150 hours =q3=8/27.
(3)Probability that a tube will last less than 200 hours given that the tube is
P(150< X <200)
still functioning after 150 hours is P(X<200|X>150)=
=
P(X > 150)
200
dx
100
x2
150
dx
100
x2
1 3
= X
6 2 =0.25.
150
Example 4.10
The amount of bread (in hundreds of pounds) x that a certain
bakery is able to sell in a day is found to be a numerical valued random
phenomenon with a probability function specified by the p.d.f. f(x) given by
f(x)=kx, 0 x<5; =k(10-x), 5 x<10; =0, otherwise. (1) find the value of k such
42
that f(x) is a p.d.f., (2)what is the probability that the number of pounds of bread
that will be sold tomorrow is (a) more than 500 pounds, (b) less than 500 pounds
and (c) between 250 and 750 pounds? (3) Denoting by A,B,C the events that the
pounds of bread sold are as in (2)(a),(2)(b) and (2)(c) respectively, find P(A|B),P(A|
C). Are (1) A,B independent, (2) A,C independent?
(1)
f ( x ) dx
=1 gives k=1/25.
10
(2)(a)P(5 X 10)=
1
( 10x ) dx
25
5
=0.5
1
xdx=
25
(b)P(0
X <5 =
0.5
7.5
1
1
(c)P(2.5 X 7.5 = 25 x dx + 25 (10x ) dx =3/4
2.5
5
(3)From (2)(a),(b),(c), P(A)=0.5, P(B)=0.5, P(C)=3/4. The events A B and
A C
1
C)=
1 ( 10x ) dx =3/8.
25 5 25
P(A)P(C)=1/2 X =3/8=P(A C ), P(A)P(B)=1/4 0=, B are not independ
P(A
B). Thus A,C are independent and A,B are not independent.
Expectation of a r.v.
Let the r.v. X take values x1,,xn with probabilities p1,,pn. Let X take value xi , fi
number
of
times;
let
N=f1++fn.
Mean
of
X
is
given
by
f 1 x 1 ++ f n x n f 1
f
= x 1 ++ n x n . Let N . Using the statistical definition of
N
N
N
probability, limiting value of mean of X ,
X =p 1 x 1+ + pn xn .
pi xi
1
43
Thus E(X) may be regarded as the limiting value of the average value of X realized
in N random experiments as N . Generalising, if f(X) is a function of X, f(X)
will take values f(x1),,f(xn) with frequencies f1,,fn and the average value of f(X)
f1
fn
in N experiments is N f(x1)++ N f(xn) and as N , this approaches to
E(f(X))=p1f(x1)++pnf(xn). In particular,
(X X )
(X X )
r (a) =E[(X-a)r],
r
r=E ], 2 =E
]=
2 .
Example 4.11
What is the expectation of the number of failures preceding the
first success in an infinite series of independent trials with constant probability p of
success in a trial?
If X denotes the number of failures preceding the first success , we find that
X takes the values 0,1,2,3, with probabilities p,qp,q 2p,q3p,, where q=1-p.
Thus probability density function is f(x)=q rp, r=0,1,2,. Hence
E(X)=0.p+1.qp+2.q2p+3.q3p+=pq(1+2q+3q2+)=pq(1-q)-2
(since
q<1)=q/p = 1/p-1.
Properties of Expectation
1. Addition Theorem of Expectation: If X,Y are r.v., then E(X+Y)=E(X)+E(Y).
2. Multiplication Theorem of Expectation: If X,Y are independent r.v.,
E(XY)=E(X)E(Y).
3. If X is a r.v. and a,b are constants, then E(aX+b)=aE(X)+b, provided all the
expectations exist.
4. If X 0,then E(X) 0.
5. If X,Y are r.v. and X(t) Y(t), forall t, then E(X) E(Y), provided all
expectations exist.
Example 4.12 Let X be a r.v. with the following probability distribution:
x:
-3
P(X=x):
1/6
1/3
Find E(X) and E(X2) and using laws of expectation, evaluate E(2X+1)2.
E(X)=
xp ( x)
1
1
1
11
=(-3) 6 +6 2 + 9 3 = 2 , E(X2)=
E(2X+1)2=4E(X2)+4E(X)+1=209.
44
93
x 2 p ( x )= 2
. Then
Example 4.13 Two unbiased dice are thrown. Find the expected values of the
sum of numbers of points on them.
The probability function of X(sum of number of heads on two dice) is
x:
2
3
4
12
P(X=x):
1/36
2/36
3/36
1/36
1
1
E(X)= px= 36 ( 2+6+ 12+ 20+30+42+ 40+36+30+22+12 ) = 36 252 =7.
Chapter V
Special Probability Distributions
Binomial Distribution
Let a series of n trials be performed in which occurance of an event is called a
success and its a non-occurrence is called a failure.Let p be the probability of a
success and q=1-p is the probability of a failure. We shall assume that trials are
independent and probability p of success is same in every trial. The number of
successes in n trials may be 0,1,2,,n and is a randam variate. The probability of x
succeses and n-x failures in a series of n independent trials in a specified
order(say) SSFSFFF FSF (S represents success and F represents failure) is given
by compound probability theorem: P(SSFSFFF FSF)=P(S)P(F)=p..p(x factors)q
n
q(n-x factors)=pxqn-x. But x successes in n trials can occur in C x ways and the
probability for each one of these ways are same, viz. p xqn-x. Hence the probability
of x successes in n trials in any order is given by the addition theorem of
n
probability by the expression C x pxqn-x. The probability distribution of the number
of successes so obtained is called Binomial probability distribution, for the
n
n
obvious reason that the probabilities of 0,1,,n successes viz. q n, C1 p1qn-1, C2
p2qn-2,,pn are the successive terms of the binomial expansion of (q+p)n.
Definition A random variable X is said to follow binomial distribution with
parameters n and p, written as X B ( n , p ) , if it assumes only non-negative values
45
and
its
p.m.f.
is
given
P(X=x)=p(x)= C x pxqn-x,
by
x=0,1,,n,
q=1-p;
Example 5.1
heads.
Ten coins are tossed. Find the probability of getting at least seven
1
2
10 x
( )( )
getting
1
2
10
( ) [C
at
10
7
least
10
10
+ C10
8 +C 9 +C 10 ]=
10 1
= Cx 2
heads
10
()
is
p(7)+p(8)+p(9)+p(10)=
176
1024 .
Example 5.2
A and B play a game in which their chances of winning are in the
ratio 3:2. Find As chance of winning at least three games out of the five games
played.
Let p be the probability that A wins the game. N=5,p=3/5, q=2/5. The
probability that out of 5 games played, A wins x games is given by:P(X=x)=
C5x
3
5
2
5
5 x
( )( )
, x=0,1,,5.
The required probability that A wins at least three games is given by P(X
5
C 5r
3)=
r =3
3r 25r
= 0.68.
55
Example 5.3
A multiple-choice test consists of 8 questions with 3 answers to
each question(of which only one is correct). A student answers each question by
rolling a balanced die and checking the first answer if he gets 1 or 2, the second
answer if he gets 3 or 4 and the third answer if he gets 5 or 6. To get a distinction,
46
the syudent must secure at least 75% correct answers. If there is no negative
marking, what is the probability that the student secures a distinction?
Since there are three answers to each question, out of which only one is
correct, the probability of getting a correct answer to a question is p=1/3,so
that q=2/3. The probability of getting r correct answers in a8-question set is
P(X=x)=p(x)=
1
C
3
8
x
2
3
8 x
( )( )
, x=0,1,,8.
Hence the required probability of securing a distinction (that is, to get correct
answers to at least 6 out of 8 questions) is given by: p96)+p(7)+p(8)=
6
2 86 8 1 7 2 87 8 1 8 2 88
8 1
C6
+C7
+C 8
=0.0197.
3 3
3 3
3 3
( )( )
( )( )
( )( )
Example 5.4 The probability of a man hitting a target is . (1) if he fires 7 times
, what is the probability of his hitting the target at least twice? (2) How many times
must he fire so that the probability of his hitting the target at least once is greater
than 2/3?
p=probability of the man hitting the target =1/4, q=1-p=3/4.
x
3 7 x
7 1
C
p(x)=probability of getting x hits in 7 shots= x 4 4
, x=0,1,,7.
( )( )
7 1
(1)Probability of at least 2 hits=1-[p(0)+p(1)]=1-[ C0 4
3
4
7 0
( )( )
+C 71
1
4
3
4
71
( )( )
4547
]= 8192 .
3
4
()
3
4
()
2
> 3
, that is,
1
3 >
3
4
. It is required to
()
. Taking logarithms
0.4771
on both sides and simplifying, n> 0.1250 =3.8. Thus required number of
shots is 4.
Example 5.5
In a Binomial distribution consisting of 5 independent trials,
probabilities of 1 and 2 successes are 0.4096 and 0.2048 respectively. Find the
parameter p of the distribution.
47
Dividing, we get
5 (1 p)
= 2, p=0.2.
10 p
x=0
x1 n x
1=E ( X ) = x Cnx p x q nx =np C n1
q =np( q+ p)n1=np .
x1 p
(central)=npq,
2 =n(n-1)p2+np,
=npq(q-p).
1
3
()
=0.99863.
Poisson Distribution
Poisson Distribution is a limiting case of Binomial Distribution under the following
conditions:
48
Definition
e x
x ! , x=0,1,
,>0; =0,otherwise.
is known as the parameter of the distribution; we write X~P() to denote X is a
Poisson variate with parameter .
Following are some examples of Poisson variate:
2= E(X2)= 2+ , 2 = 2( 1)2 = ,
= .
Example 5.7
Find the probability that at most 5 defective fuses will be found in
a box of 200 fuses if experience shows that 2 percent of such fuses are defective.
n=200, p=probability of defective fuses=2%=.02. Since p is small, we may
use oisson distribution. =mean number of defective pins=np=200(.02)=4.
5
e x !4
x=0
=e-4
4 4 4
4
1+ 4+ + + +
2 6 24 120
=.785.
Example 5.8
Six coins are tossed 6400 times. Using Poisson distribution, find
the approximate probability of getting six heads r times.
The probability of getting six heads in one throw of six coins (a single trial) is
p=
1
2
()
1
2
()
P(X=r)=
100
r!
, r=0,1,2,
49
Example 5.9
In a book of 520 pages, 390 typographical errors occur. Assuming
Poisson law for the number of errors per page, find the probability that a random
sample of 5 pages will contain no error.
The average number of typographical error per page in the book is
=390/520=0.75.
Using Poisson probability law,the probability of x errors per page is given by
x
e0.75 (0.75) x
e
P(X=x)= x ! =
, x=0,1,2,. The required probability that a
x!
random sample of 5 pages will contain no error is given by [P(X=0)] 5=(e0.75 5
) =e-3.75.
Normal Distribution
Definition A r.v. X is said to have a normal distribution with parameters
(called mean) and
{ ( )}
1
1 x
law: f(x; , = 2 exp 2
).
If X~N( ,
), then Z=
variate
is
given
1
( z )=
e
2
by
z
2
,< z < .
z
The
z
(u ) du= 12 e
distribution function, denoted by (z)=P(Z z =
corresponding
u
2
du .
(-z)=1- (z)
P(a X b =
( b
)
( a
)
50
, where X~ N( ,
).
and s.d.
is given by f(x)=
1
e
2
( x )
2
2
occurring at x= .
1=0, 2 =3
Since f(x) (being probability) 0, for all x, no portion of the curve lies
Area
property:
P(
1 (about
X=10)=40.
mean=10+ 1 =50.
Also
4 (about
=3
Thus
=48 giving
=2.
51
2012
=2. Thus P(X 20)=P(Z 2)=0.5-P(0 Z 2
4
=0.5-0.4772=0.0228.
P(X 20)=1- P(X 20)=1-.0228=.9722.
P(0 X 12 =P(-3 Z 0)=P(0 Z 3 =0.49865.
x12
(b)When
X=x,
Z=
=z1(say).GivenP(X>x)=P(Z>z1)=0.24;thus
4
P(0<Z<z1)=0.26. From normal table, z1=0.71(approx..) Hence
x12
4
),
with
=662 ,
=32.
(1)
The probability that a plot has a yield over 700 kilos is given by
P(X>700)=P(Z>1.19) [z=
700662
=1.19
32
= 0.5-P(0 Z 1.19)=0.5-
650662
] =1000 x P(Z>0.38)=1000
32
X [0.5-P(0 Z 0.38)]=1000[0.5-0.1480]=352.
(3)
100
The lowest yield , say, x,of best 100 plots is given by:P(X>x)= 1000
x662
=0.1. When X=x, Z=
=z1 (say) such that P(Z>z1)=0.1 P(0 Z
32
z
)=0.4z1=1.28(approx.)
(from
normal
tables).
Thus
x=662+32z1=702.96. Hence the best 100 plots have yield over 702.96
kilos.
52
Example 5.13
The marks obtained by a number of students for a certain
subject are assumed to be approximately normally distributed with mean value 65
and s.d. 5. If 3 students are taken at random from this list, what is the probability
that exactly 2 of them will have marks over 70?
Let the r.v. X denote the marks obtained by the given set of students in the
=65 ,
=5. The
given subject. Given that X~ N( , 2), with
probability that a randomly selected student from the given set gets marks
over 70 is given by p=P(X>70)=P(Z>1)=0.5-P(0 Z 1 =0.50.3413=0.1587. Since this probability is same for each student of the set,
the required probability that out of 3 students selected at random from the
set, exactly 2 will have marks over 70, is given by the binomial probability
law:
Chapter VI
Correlation and Regression
Often we come across situations in which our focus is simultaneously on two or
more possibly related variables . If change in one variable affects a change in the
other variable, the variables are said to be correlated. If increase in values of one
variable results in increase in the corresponding values of the other variable,
variables are said to be positively correlated; if increase in one variable results in
decrement of values of the other variable, variables are negatively correlated.
Correlation is said to be perfect if deviation in one variable is followed by a
corresponding and proportional deviation in the other variable.
Scatter Diagram
It is simplest diagrammatic representation of bivariate data. Thus for the bivariate
distribution (xi,yi), i=1,,n; if the values of the variables X and Y are plotted along
the x-axis and y-axis respectively in the x-y plane, the diagram of dots so obtained
is known as scatter diagram. From the scatter diagram, we can form a fairly good,
53
though vague, idea whether the variables are correlated or not: if the points are
very close to each other, we should expect high correlation between the variables.
Karl Pearsons coefficients of correlation
Let (xi,yi),i=1,,n be a bivariate distribution of two r.v.s X and Y. Correlation
coefficient between X,Y , usually denoted by r(X,Y) or by r XY, is a numerical
measure of linear relationship between them and is defined as : r(X,Y)=
n
Cov ( X , Y )
1
( xi x )( y i y ) =
,
where
Cov(X,Y)=E[{X-E(X)}{Y-E(Y)}]= n
X Y
1
1
1
1
x i y i y x i
x yi
- n
+ x y
n
n
2X =
1
x y
x y ,
n i i -
1
1
1 1
1
xi x ) 2= x 2i x 2
2Y = = ( y i y )2= y 2i y 2
(
,
.
n
n
n n
n
There is a linear relationship between the r.v.s, that is, if the paired
observations of both the variables are plotted on a scatter diagram, the
plotted points will approximately be concurrent
The variations in the two variables follow the normal law.
1
( x ix )( y i y )
n 1
( ai bi )
Cov (X , Y )
2
=
r =
x ,
1
2
2
r(X,Y)=
X Y
ai )( bi ) , where ai= xi1
21
2 2
(
( x ix ) n ( y i y )
n
( ai b i ) ( a2i )( b2i )
. Hence r2 1.
Thus -1 r 1 .
Note if r=+1, correlation is perfect and positive; if r=-1, correlation is perfect and
negative.
Effect of change of origin and scale of reference on correlation coefficient
54
Xa
Y b
,V=
If U=
h
k
then
r XY =r UV
, then
r XY =
hk
r UV
. Hence if h,k are of the same sign,
h2 k 2
r XY =r UV
Cov ( X , Y )
=0. Two
X Y
65
66
67
67
68
69
70
72
Y:
67
68
65
68
72
72
69
71
55
Cov (X , Y )
r(X,Y)=
=
X Y
1
XY X Y
n
1
1
X 2 X 2
Y 2Y 2
n
n
)(
) {
1
X 3756068 X 69
8
}{
37028
38132
682
692
8
8
=0.603.
Short-cut Method
X
65
66
67
67
68
69
70
72
TOTAL
67
68
65
68
72
72
69
71
U=X68
-3
-2
-1
-1
0
1
2
4
0
V=Y69
-2
-1
-4
-1
3
3
0
2
0
U2
V2
UV
9
4
1
1
0
1
4
16
36
4
1
16
1
9
9
0
4
44
6
2
4
1
0
3
0
8
24
1
1
UV U V
X 24
U
=0, V =0, Cov(U,V)= n
= 8
=3,
36=4.5,
2V =
1
1
V 2V 2
X
=
44=5.5.
n
8
Thus
rUV=
2U =
1
1
U 2U 2
X
= 8
n
Cov (U , V )
=
U V
3
4.5 X 5.5
=0.603.
Example 6.2
A computer while calculating correlation coefficient between two
variables X and Y from 25 pairs of observations obtained the following results:
2
2
n=25, X =125, X =650, Y =100, Y =460, XY =508.
It was however later
discovered at the time of checking that he had copied down two pairs as (6,14),
(9,6) while the correct values are (8,12),(6,8). Obtain the correct value of
correlation coefficient.
Corrected
Corrected
X =
125-6-8+8+6=125,corrected
X2
=650-62-82+82+62=650,
Y =
100-14-6+12+8=100
corrected
62+122+82=436,
Corrected XY =508-6 X 14-8 X 6+8 X 12+6 X 8=520.
56
Y2
=460-142-
Corrected
1
X =
25
x 125=5, Corrected
1
2
XY X Y
Cov(X,Y)= n
=4/5. X =1,
1
Y =
25
x100=4.
2Y =36/25.
Hence
corrected
rXY=0.67.
Regression Analysis
Regression Analysis is a mathematical measure of the average relationship
between two or more variables in terms of the original units of data.
If the variables in a bivariate distribution are related, the corresponding points in
the scatter diagram will cluster round some curve called curve of regression. If
the curve is a straight line, it is called line of regression and there is said to be
linear regression between the variables.
The line of regression is the line which gives the best estimate to the value of one
variable for any specific value of the other variable. Thus line of regression is the
line of best fit and is obtained by principle of least squares.
Let us suppose that in the bivariate distribution (x i,yi), i=1,,n, X is independent
and Y is dependent variable. Let the line of regression of Y on X be Y=a+bX (1).
(1) represents a family of straight lines for different values of a and b. The
problem is to find a and b corresponding to the line of best fit.
According to the principle of least squares, we have to find a,b so that E=
n
( y i ab xi )2
1
( y iab x i)
n
E
=2
a
1
n
y i=na+
which gives
xi
1
(2),
x i yi =a x i+ b x i (3)
1
1
1
2
Equations (2) and (3) are known as normal equations for estimating a and b.
From (2), on dividing by n, we get y =a+b x (4) Thus the line of regression of Y
on X passes through ( x , y ).
57
1
x y x y
Now A(say)=Cov(X,Y)= n i i
2X =
1
x 2 x 2
n i
1
x y
n i i =A+ x y . (5)
x 2i = 2X + x2
1
.(6)
n
2
2
2
From (3),(5) and (6), A+ x y =a x +b( X + x ). From (4) and (6), A=b X
2
giving b=A/ X .
Since the regression line of Y on X passes through ( x , y ) and has slope b=A/
2X , its equation is
Y
y = A / 2X
x
y
=
Y(X), that is, Yr X (X- x ).
X
x
Note In case of perfect correlation r= 1 and in that case the equations of two
regression lines coincide:
Y y
X x
Y =
X .
Regression Coefficients
Y
bYX= r X
X
and bXY= r Y
Y respectively.
Properties of regression coefficients
A
A
A
bYX bXY=r2. Thus r= b XY bYX . Since r= X Y , bYX= 2X , bXY= 2Y , it may
be noted that sign of correlation coefficient is same as that of regression
coefficients, since the sign of each depends on that of A. Thus, if the
58
If one of the regression coefficients is >1, then the other must be <1: b YX
bXY=r
1
1; if b =
b XY >1, then bXY<1.
YX
Example 6.3 Obtain the equations of two lines of regressions for the following
data. Also obtain the estimate of X for Y=70.
X:
65
66
67
67
68
69
70
72
Y:
67
68
65
68
72
72
69
71
Let
U=X-68,
V=Y-69.
Then
2
2
U=0,
V =0 , U =4.5, V = 5.5,
= V
X =68+ U
=68, Y =69+ V =69, X = U = 4.5 =2.12, Y
=
5.5 =2.35.
Y
X
Equation of line of regression of X on Y is: X=r Y (Y- Y ), or,
X=0.54Y+30.74
To estimate X for given Y, we use line of regression of X on Y.If Y=70,
^
estimated value of X is given by X =0.54 X 70+30.74=68.54.
Example 6.4 In a partially destroyed laboratory, record of an analysis of
correlation data, the following results only are available:
Variance of X=9,
59
Regression equations: 8X-10Y+66=0,40X-18Y=214. What are values of(1) X , Y ,
(2)rXY, (3)
(1)Since ( X , Y ) is the point of intersection of the lines of regression,
solving given equations of lines of regression simultaneously, we get
=13, Y =17.
8
66
X+
(2) Comparing given equations of regression lines Y= 10
10 , X=
18
214
8 4
18 9
9
Y+
=
=
2
40
10 , we get bYX= 10 5 , bXY= 40 20 . Hence r = bYX. bXY= 25 .
Hence r=
3
3
5 . Since both the regression coefficients are positive, r= 5
=0.6.
Y
(3)we have bYX=r. X
; hence
4
3
5 = 5
Y
3 , giving
=4.
Example 6.5 Find the most likely price in Mumbai corresponding to the price of
Rs. 70 at Kolkata from the following:
Kolkata
Mumbai
Average price
65
67
Standard Deviation
2.5
3.5
Correlation coefficient between the prices of commodities in the two cities is 0.8.
Let the prices (in Rs.) in Kolkata and Mumbai be denoted by X and Y
Y
Line of regression of Y on X is: Y= r. X (X- X ) ,or, Y=67+0.8 X
3.5
2.5 (X-65).
60
When X=70,
Y^ =67+0.8 X
3.5
2.5 (70-65)=72.6.
that E=
E
2
x i ( y iab1 x ib2 x i ) ,
0= b 1 =2
1
( y iab1 x ib2 xi ) ,
1
2
0=
E
=2 x 2i ( y iab1 x ib2 x 2i ) .
b 2
1
y i=na+b1 x i+ b2 x2i
Simplifying,
x 2i y i =a x 2i + b1 x 3i +b 2 x 4i
x i yi =a x i+ b1 x2i +b 2 x 3i
Additional units(Y): 2
10
12
10
14
11
14
61
1
1
2
2
3
3
4
5
6
7
34
7
7
10
8
12
10
14
11
14
95
X2
X3
X4
XY
X2Y
1
1
4
4
9
9
16
25
36
49
154
1
1
8
8
27
27
64
125
216
343
820
1
1
16
16
81
81
256
625
1296
2401
4774
2
7
14
20
24
36
40
70
66
98
377
2
7
28
40
72
108
160
350
396
686
1849
Corresponding
normal
equations
are:
10a+34b1+154b2=95,34a+154b1+820b2=377,
154a+820b1+4774b2=1849.
Solving , a=1.80,b1=3.48,b2=-0.27. Thus regression equation of Y on X is:
Y=1.80+3.48X-0.27X2.
Chapter VII
Index Numbers
An index number may be defined as a measure of the average change in a group
of related variables over two different situations. The group of variables may be
the prices of a specified set of commodities, the volumes of production in different
sectors of an industry, the marks obtained by a student in different subjects and
so on. The two different situations may be either two different times or two
different places.
The most commonly used index number is the index number of prices. Let p 0 and
p1 denote the prices of a commodity in suitable units in two different situations
denoted by 0and 1. Any change in the price of the commodity from 0 to 1 may
be expressed either in absolute or relative terms. The absolute change is p 1-p0;
the relative change is given by p 1/p0, which is called a price relative. The problem
is to combine these various individual changes in prices and get a measure of the
62
overall change in the prices of the set of commodities. A price index number is a
sort of average of these individual price relatives, and it measures the price
changes of all the commodities collectively.
Although different commodities may have peculiar characteristics in their price
fluctuations, it has been empirically found that , taken as a whole, the distribution
of price relatives is bell-shaped with a marked central tendency, provided the base
period is in the recent past. Hence we are justified in taking an appropriate
measure of central tendency in combining the different price relatives.
Let us denote by p0i the price of i th commodity in the base period and by p 1i the
price of this commodity in the current period (i=1,,k). If we use the arithmetic
p 1i / p0 i
i
mean of price relatives for constructing the index number, then I 01=
is
k
a simple or unweighted index number.
Choice of weights
The commodities included in the index number are not all of equal importance. For
instance, in constructing a wholesale price index for India, rice should have
greater importance than tobacco. So the problem of weighting different
commodities included in the index number according to their importance deserves
attention. If we ignore weights, we get an inappropriately weighted index. If w i be
the weight attached to the price relative for the i th commodity, then we get the
weighted A.M.
p
I01=
p1 i w i
0i
wi
numbers:
Choosing wi=q0i (the base period quantities) we get Laspeyres index: I01=
p 1i q 0i
i
p 0i q 0 i
63
Choosing wi=q1i (the current period quantities) we get Paasches index: I01=
p 1i q 1i
i
p 0i q 1i
i
Choosing
wi=(q1i+q0i)/2,
we
get
Edgeworth-Marshall
index:
I01=
p 1i (q1 i+ q0 i )
i
p 0i (q1 i +q0 i )
i
p1 i q0 i p1 i q1 i
i
p0 i q0 i p0 i q1 i
i
Example 7.1 Table below gives the wholesale prices (p) and quantities produced
(q) of a number of commodities in Delhi. Calculate Laspeyres, Paasches,
Edgeworth-Marshall and Fishers index numbers for the year 1985 , with the year
1982 as base.
Commodity
1982
p
1985
q
Rice
277.92
1.1
366.67
6.2
Wheat
176.25
106.0
186.58
116.9
Jowar
151.00
4.2
182.57
5.5
Barley
121.83
2.4
181.25
1.0
Bajra
156.75
13.1
155.75
6.1
Gram
273.00
1.0
498.83
0.6
Let p0i,q0i and p1i,q1i denote the prices and quantities for 1982 and
1985, respectively. Then
p0 i q0 i
i
=22241.229,
p1 i q0 i
i
=23921.766,
6519.314.
64
p0 i q1 i
i
=24399.034,
p1 i q1 i
i
=2
23921.766
x 100
Thus Laspeyres Inex= 22241.229
=107.56
26519.314
x 100
Paasches Index= 24399.034
=108.69
50441.08
x 100
Edgeworth-Marshall Index= 46640.263
=108.15
and Fishers ideal index number=
Chapter VIII
Analysis of Time Series
Time series is a series of observations recorded at different points or intervals of
time. Maximum temperature of a place for different days of a month, yearly
production of coal for last 20 years, monthly sales figure of some product are
examples of time series data.
Let yt denotes the value of the variable y at time t (t=1,..,n).In case the figures
relate to n successive periods (and not points of time), t is to be taken as the midpoint of the t th period.
Components of time series
A graphical representation of a time series shows continual change over time,
giving us an overall impression of haphazard movement. A critical study of the
series will, however, reveal that the change is not totally haphazard and a part of
it, at least, can be accounted for. The systematic part which can be accounted for
may be attributed to several broad factors: (1)secular trend, (2)seasonal
variation, (3)cyclical variation. Separation of the different components of a
time series is of importance, because it may be that we are interested in a
particular component of the systematic variation or that we want to study the
series after eliminating the effect of a particular component. It may be noted that
it is the systematic part of the time series which may be used in forecasting.
65
observations. Each of these means is centered against the time which is the midpoint of the time interval included in the calculation of the moving average. Thus
when k, the period of moving average, is odd, the moving average values
correspond to tabulated time values for which the time series is given. When k is
even,the moving average falls midway between two tabulated values. In this case,
we calculate a subsequent two-item moving average to make the resulting moving
average values correspond to the tabulated time periods.
The interpolation ofsimple moving averages is very simple. A k-point moving
average may be interpreted as the estimated value for the middle of the period
covered from successive linear curves fitted through the first k points, through the
2nd to the (k+1)th values and so on, and lastly through the last k points.
Consider the first k points y1,,yk. Let the origin be shifted to the middle of the
t
period so that i =0. The normal equations for fitting a curve Y=a+bt through
y1,,yk are
yi
=ka+b
So that
a^ =
ti
t i y i =a ti + b t2i
t y
y i = y , b=
^ i i
k
t2i
^
period covered ,that is, for t=0, from the curve Y= a^ + b t is
a^ , which is the
first moving average value. Similarly it can be shown that the estimated value
k+1
1
y
k i=2 i , the second
year
Output(in
tons)
1992
1993
1994
1995
1996
1997
1998
1999
1542
1447
1552
2102
2612
3195
3537
3567
000 3-yearly
moving total
..
4541
5101
6266
7909
9344
10299
..
3-yearly
moving
average(Trend
values)
..
1513.7
1700.3
2088.7
2636.3
3114.7
3433.0
..
Example 8.2 Work out the trend values by 4-yearly moving average from the
following data on production of iron ore( in 000 tons)
Year:
1983 84
85
86
87
88
89
90
91
92
93
Production:
110
118
134
121
132
145
155
159
148
162
Year
1983
125
production 4-yearly
moving total
110
1984
125
1985
118
487
1986
121.75
123.125
498
124.5
505
126.25
134
1987
121
1988
132
125.370
129.620
532
1989
4-yearly
4-yearly(centered)
moving average moving average
133.00
135.625
553
138.25
591
147.75
145
1990
155
1991
159
143.000
149.750
607
151.75
153.870
624
1992
148
1993
162
156.00
68
Example 8.3 Calculate the trend values by the method of moving averages from
the following data on quarterly production(in 000 tons):
year
Quarter
1995
1996
1997
15
15
20
II
19
22
21
III
21
23
25
IV
18
20
20
Is it possible to find the trend value for the first quarter of 1998 by the above
method? Justify.
Year Quarter Production
4-Quarter
1995
4-Quarter
moving average
4-Quarter(centered)
moving average
15
----
II
19
---73
III
18.25
21
18.250
73
IV
18.25
18
18.625
76
15
19.00
78
1996
II
19.250
22
19.50
80
III
19.750
23
20.00
85
20.625
69
IV
20
21.25
84
21.125
20
21.00
86
1997
II
21.250
21
21.50
86
III
21.500
25
21.50
----
IV
20
-----
It is not possible to find the trend values for the 1 st quarter of 1998 by the moving
average method since there is no specific mathematical equation which can be
used for interpolation or prediction purposes.
Method of mathematical curves
The trend values obtained by the method of moving averages , even though fairly
smooth, is not representable by a simple mathematical formula.Since there does
not exist any mathematically expressed trend equation, the method fails to
achieve the main objective of trend analysis, that is, the interpolation and
extrapolation of trend values.Therefore, attempt is made to fit the observed time
series with a fairly simple mathematical curve. The fitting of mathematical curve
has two parts: (1) determination of the appropriate trend curve, (2) determination
of unknown parameters involved in the equation. From the graphical
representation of the given time series , an investigator may guess the nature of
the which fits the data best. The method is subjective in this sense. Determination
of unknown constants appearing in the trend equation can be done by method of
least squares.
Example 8.4 Following table gives the number of hospital beds in West Bengal
for the years 1979 to 1986. Plot of year versus no. of beds suggest that a linear
trend will be appropriate to fit to the given data. The necessary data are done in
table below:
Year
1979
1980
No.of
beds
55477
58045
t=2(yearmid-period)
-7
-5
tyt
t2
Tt=a0+a1t
-388339
-290225
49
25
55938
57365
70
1981
1982
1983
1984
1985
1986
Total
58448
59876
61894
63734
64667
65319
487460
Since
-3
-1
1
3
5
7
0
-175344
-59876
61894
191202
323335
457233
119880
9
1
1
9
25
49
168
58792
60219
61646
63073
64500
65927
a0=60932.5,
a1=713.57.
Tt=60932.5+713.57t.
The
linear
trend
equation
is
therefore,
Pass-Kilo
1983
1984
1985
1986
1987
1988
1989
Total
6096
6379
6774
7327
7516
7863
8427
50382
Here
0=
=0,
t3
T=year1986
-3
-2
-1
0
1
2
3
0
tyt
t2yt
t2
t4
-18288
-12758
-6774
0
7516
15726
25281
10703
54864
25516
6774
0
7516
31452
75843
201965
9
4
1
0
1
4
9
28
81
16
1
0
1
16
81
196
yt
=7a0+28a1, 10703=
a0=7176.63,
a1=5.20,
Tt=7176.63+5.20t+382.25t2.
=a2
a2=382.25.
t2
, 201965=28a0+196a1. Solving,
Thus
trend
T=year-1986
-3
-2
-1
0
1
2
3
a2t
-1146.75
-764.5
-382.25
0
382.25
764.5
1146.75
a1t2
46.80
20.80
5.20
0
5.20
20.80
46.80
Trend Tt=a0+a2t+a1t2
6076.68
6432.93
6799.58
7176.63
7564.08
7961.93
8370.18
71
equation
is
72