Introduction to statistics and

probability
Prepared by Ranju Mohan, PhD student, CE dept, IITM
For CE302 class by Gitakrishnan Ramadurai, AP, CE dept, IITM
Statistics:

Mean, Median,
Mode etc

Standard deviation,
Variance, Range,
Quartiles etc
Working with data…..

Sample + Descriptive statistics

(Measures of spread & Measures of
dispersion)

Statistical estimate
POPULATION +
Inferential statistics
Sample Concepts of
probability

INFERENCE about the

population
 Descriptive statistics
Speed data collected for 10 vehicles:
Speed 32.4  3  27.8  43.5  52.3  36.9  29.5  49.8  58.9
(km/hr.) Mean 
10
32.4
27.8  39.59
27.8
29.5
43.5
32.4 Arranging in increasing order,
52.3
32.4
36.9
32.4 32.4  36.9
Median   34.65
29.5
36.9 2
32.4
43.5
49.8
Mode = 32.4
58.9
52.3
Measures of Central
32.4
58.9
Tendency
Descriptive statistics – Measures of spread

Speed
(km/hr.) Range = Max value –Min value = 58.9- 27.8 = 31.1
32.4
27.8
27.8
29.5 Absolute deviation: |value - mean|
43.5
32.4 Ex: Absolute deviation = |52.3- 39.59|=12.71
52.3
32.4 Arranging in increasing order,
36.9
32.4
Quartiles 1st (25% of 2nd (25% of 3rd (25% of 4th (25%
29.5
36.9 data) data) data) of data)
32.4
43.5 Value [1(10)+2]/4th [2(10)+2]/4th [3(10)+2]/4th -
49.8
49.8
58.9 = 32.4 (Q1) =34.65 (Q2) =49.8 (Q3)
52.3
32.4 Interquartile range = Q3-Q1 = 17.4
58.9
Descriptive statistics - Measures of spread
Speed Variance :
(km/hr.)
n
 xi  x  2

32.4
(32.4 – 39.59)2
27.8
s2  = 684.21/9 = 76.02
(27.8 – 39.59)2
i 1 n 1
43.5
(43.5 – 39.59)2
52.3
(52.3 – 39.59)2
36.9
(36.9 – 39.59)2 Standard deviation :
29.5
(29.5 – 39.59)2
n
 xi  x  2
32.4
(32.4 – 39.59)2
49.8
(49.8 – 39.59)2
s 
i 1 n 1
= √684.21/9 = √76.02 = 8.72

58.9
(58.9 – 39.59)2
32.4
(32.4 – 39.59)2
∑ = 684.21
Data:
 Groups of information that represents the qualitative or
quantitative attributes of a variable or a set of variables.

 Visual/Graphical Representation:
 Frequency distributions
 Graphs
 Box plot
 Scatter plot
 Stem and leaf
 Data representation: Examples
Bar chart
Year No. of Accidents

Total no. of
Fatal Non- Total accidents
fatal
2002 4 7 11 Graphical
representation
2003 7 6 13
2004 4 6 10
Year
2005 4 4 8
2006 2 4 6
2007 3 3 6 No. of
accidents

Frequency distribution table

 Data representation: Examples
Previous data ,

Year No. of
Accidents
2002 11
2003 13
2004 10
2005 8
2006 6
2007 6 Frequency polygon
Data representation: Examples

Pie Diagram

Daily volume of vehicles observed :

Vehicle type frequency
Bus 35 Vehicle
composition
Car/Jeep 160
Truck 22
Bicycle 40
Others 25
TOTAL 282
 Data representation: Examples
Given speed data:
63.2 49.9 36.9 44.2 54.8 49 42.9 32.4
54.3 37.5 45.1 51.7 47.5 43.8 55.9 48.8
41.1 47.5 52.3 39.2 57.3 36.3 42.8 58.7
52.9 42.5 46.4 53.3 46.5 43.2 56.9 47.7
47.8 35.6 50.3 44.7 46.2 38.4 62.4 39.4
56.4 55.1 64.8 52.8

3 2.4 5.6 6.3 6.9 7.5 8.4 9.2 9.4

4 1.1 2.5 2.8 2.9 3.2 3.8 4.2 4.7 5.1 6.2 6.4 6.5 7.5 7.5 7.7 7.8 8.8 9 9.9

5 0.3 1.7 2.3 2.8 2.9 3.3 4.3 4.8 5.1 5.9 6.4 6.9 7.3 8.7

6 2.4 3.2 4.8

57.3
Stem and Leaf plot
 Data representation: Examples
Given speed data (km/hr.),
63.2 49.9 36.9 44.2 54.8 49 42.9 32.4
54.3 37.5 45.1 51.7 47.5 43.8 55.9 48.8
41.1 47.5 52.3 39.2 57.3 36.3 42.8 58.7
52.9 42.5 46.4 53.3 46.5 43.2 56.9 47.7 Speed class No. of
47.8 35.6 50.3 44.7 46.2 38.4 62.4 49.4 vehicles
56.4 55.1 64.8 52.8 30-35 1
35-40 6
Group into different speed class 40-45 8

max value  min value 45-50 12

Class Interval 
1  3.22 log  No. of veh 50-55 8
64.8 - 32.4 55-60 6

1  3.22 log(48) 60-65 3
 5.05, say 5
Data representation: Examples

Speed No. of
class vehicles
30-35 1

No. of vehicles
35-40 6
40-45 8
45-50 12
50-55 8
55-60 6
60-65 3 30 35 40 45 50 55 60 65
Speed (km/hr.)

Histogram
 Data representation: Examples

Speed No. of Cumulative

class vehicles no. of veh.
30-35 1 1
35-40 6 7
40-45 8 15
45-50 12 27
50-55 8 35
55-60 6 41
60-65 3 44

Speed (km/hr.)

Ogive
Q: Number of vehicles with speed less than 50 km/hr. ?
Ans: 31
 Data representation: Examples
Paired data set:
(Number of accidents ,Vehicle speed)

Speed No. of
(km/hr.) accidents
22 1
How to relate?
45 3
32 4
75 8
66 10
58 6
Scatter plot
 Data representation: Examples

When having several simultaneous comparison: 11

observations in 3 diff. days. (n=11) Box Plot
Speed (km/hr)
Mon Wed Fri Points to be plotted Speed (km/hr.)
L 25
32 33
22 24 H Mon Wed Fri
Q3
25
28 22
29 33
29 Q3 H
Lowest (L) 25 28 28
Q1 33
65 70
30 58
31
Q3
46
38 42
33 36
33 Q1 = [1(n)+2]/4th value Q2 33 Q230 31
Q2
= 3rd value
28
39 29
40 29
36 Q1
Q1
Q2 42
59 55
42 38 L
L
L
Q2 = median 42 42 38
38
46 40
55 45
40
Q3= [3(n)+2]/4th value 59 64 45
55 60 31
41
=9th value
Q3 59
42 72
64 50
45
39
60 64
70 41
50 Highest (H) 65 72 58
Day
H 65
60 30
72 40
58
Inferential statistics- basic concepts
 Use sample statistics and probability concepts to make
inferences about the population

 Probability (P): The likelihood of something happening or

being true.

 Based on the assumption that sampling is random

Inferential statistics :
Probability concepts-Random variables

Random
Experiment

x1 x2 x3 x4 x5 Outcomes

Assigns unique
values to the
Discrete
outcomes Random
variable (a
Function)
Continuous
Probability concepts – Random variables
Discrete Random variable Continuous Random variable

Probability mass function, pmf = p(x) Probability density function, pdf = f(x)

b
p(x) = P(X=x)
 f(x)dx  P(a  x  b)
0  p(x)  1 a
f(x)  1
 p(x)  1
 f(x) 1
Cumulative distribution function F(X ≤ x) Examples:
x P(X=x) f(x) = x, 0 ≤ x ≤ √2; 0 otherwise
1 0.13
F(x) = x2/2
2 0.27 F(3) =0.13+0.27+0.25
3 0.25
4 0.15 = 0.65
5 0.20
Probability concepts – Random variables
Discrete Random variable Continuous Random variable

Given xi’s and p(xi)’ s Given xi’s and f(xi)’s

Expectation of a random variable, E(X): Weighted average of the possible values

E(X)   x ip(x i ) E(X)   xf(x)

i

E(X2 )   x i p(x i )
2 E(X2 )   x 2 f(x)
i

Mean = E(X) = First moment about origin

Variance =V(X) Second moment about mean V(X)  E(x  μ )2

or
V(X)  E(x2 )  E(x)2
Some common probability distributions used
in traffic engineering

Discrete data Continuous data

Bernoulli distribution Exponential distribution

Binomial distribution Normal distribution &

distribution arising from
Multinomial distribution normal

Poisson distribution Chi-square distribution

t- distribution

F – distribution
Special Random variables and probability
distributions
Discrete Random variables
Bernoulli :
Two possible outcomes for one trial:
‘success’ (X=1) or ‘failure’ (X=0)

P(X  0)  1- p
pmf   0  p 1
P(X  1)  p

Mean = p; Variance = p(1-p)

Binomial :
‘n’ independent trials, each having two
outcomes
n!
pmf  p(X  x )  p x qn x ; x  0,1,...,n
x!(n  x )!

Mean = np; Variance = np(1-p)

Special Random variables and probability distributions
Discrete Random variables
Poisson:
When ‘n’ is large and p is small
e   x
pmf  P(X  x)  ; i  0,1,.....
x!
  mean number of successes  np
x  actual number of successes

Mean = ν; Variance = ν
Continuous Random variables
Normal:

 ( x  ) 2
1
pdf  e 22
;-  x  
 2

Mean = µ ; Variance = ϭ2
Special Random variables and probability distributions
Continuous Random variables
Normal random variable, z

x 
z

When µ=0 and ϭ =1;
x2
1 -z 0 z
pdf  e 2
;-  x  
2
Exponential:

e  x if x  0
P( x )  
0 if x  0

Mean = 1/λ ; Variance = 1/λ2

Examples
 Que: On a particular junction, out of two routes to a particular destination,
probability of choosing 1st route is twice as that of the 2nd route. How many
number of vehicles will turn to Route 1 when a total of 5 vehicles reach at
the junction at a specified time?
p(route 1)  p x (1 p)1 x
x  0 with probability 0.33
x  1 with probability 0.67

p(route 1)  0.330 (1 0.33)1 


0.33
or   0.67
p(route 1)  0.671 (1 0.67)0 

Ans: Number of vehicles choosing Route 1

Bernoulli distribution = 0.67×5 = 3.3, say 3
 Examples
 Que: Probability of choosing a particular route is 1/5. Find out the
probabilities that out of 5 vehicles reaching that location, exactly 0, 1, 2, 3, 4, 5
vehicles will choose that particular route.

With Binomial distribution,

Ans: x P(x)
n!
P(x)  p x qn x 0 0.33
x!(n  x)!
1 0.41

n = 5 ; p =1/5 2 0.20
3 .05
4 .006
5 .0003
Examples
 Ques: For 3 different routes at a particular location, probability of choice is
given by 0.35, 0.40, 0.25 respectively. What is the probability that out of 5
vehicles reaching at the location, one, three and two vehicles will choose the
route 1, 2 and 3 respectively.

By multinomial distribution,
n! x x x
p( x1 , x 2 ,....x k )  p1 1 p 2 2 ....p k k
x1! x 2 !...x k !

5!
p(1, 3, 2)  0.351  0.43  0.252
1! 3! 2!

Ans: 0.014
Examples
 On a motorway, the number of vehicles arriving from one direction in
successive 10 sec intervals was counted and is given below. Find out the
probabilities P(0), P(˃ 3), P(3˂ X˂ 6) etc.

By Poisson distribution, No.No. of veh. in

of veh. P(x)
Frequency Total no. Total time
in 10 10
secsec
(i) (x) (ii) of veh. (ii*10)
e   x
p( x )  (i*ii)
x! 0
0
11
0.135
0 110
ν = (200/1000)*10 = 2 1 0.27
1 28 28 280
2 0.271
2 30 60 300
3 0.18
3 18 54 180
Ans: 4 0.091
4 8 32 80
5 0.036
P(0) = 0.135 5 4 20 40
6 0.012
P(X˃3) = 0.144 6 1 6 10
7 .0034
P(3˂ X< 6) =0.127 7 0 0 0
∑ = 200 ∑ =1000
Example:
 Ques: If an average of 3 trucks arrive per hour to be unloaded at a
warehouse, what are the probability that the time between the arrivals of
successive trucks will be (i) less than 5 min., (ii) at least 45 min.

Using exponential distribution,

P ( x )   e  x
x
P(X  x )    e -x dx  1- e -x
0

λ = 3/60 = 0.05 veh/min.

Ans:

P(X<5)= 1-e-0.05(5) = 0.2212

P(X≥45) = e-0.05(45) = 0.105

Example:
 The spot speed at a particular location are normally distributed with a
mean of 51.7 km/hr. and std. deviation of 8.3 km/hr. what is (ii)the
probability that speed exceeds 65 km/hr. (ii) the 85th percentile speed.
x  P(X≤65)
z
 P(X>65)

(i) z = (65-51.7)/8.3 =1.6

0 1.6
From the standard normal distribution table,
F(1.6)= P(X≤65) = 0.9452 Ans: P(X>65) = 0.0548

(ii) P(X≤x) = 0.85 = F(z)= 0.85 P(X≤x)

From the standard normal distribution table,
z= 1.04
x = 1.04(8.3)+51.7 = 60.33
0 1.04
Ans: x =60.33 km/hr.
Inferential statistics: Sampling distributions
 Sampling Theory:
If a random sample of size n is taken from a population of mean  and variance  2 ,
then the sample mean X follows a normal distributi on with mean  and variance  2 /n.

The standard error of mean is given by .
n

 Central limit theorem:

If x is the mean of a sample of size n taken from a population of mean μ and variance σ 2 ,
x-μ
then the variate z  approaches a normal distribution as n 
 
 σ 
 
 n
Central limit theorem-Error estimate

X 
1   z / 2   z / 2

n
/2 /2
X 
 z / 2
 z / 2 z / 2 
n

z  / 2
E  X   , where ' ' is th level of significance
n

 Level of significance: the probability that the computed estimate will lie
outside the indicated range .Here the range is the confidence level,1  
Example
 While determining the mean speed of veh. on a section of a road, engineer
wants to be able to assert with 95% confidence that the mean speed is off
by 2.5 km/hr. If std. deviation is 8.2 km/hr., how large the sample is?

z  1    0.95
E  X    /2
n
 / 2  .025  / 2  0.025
1    0.95;   0.05; /2  0.025

z /2  1.96  1.96 1.96

1.96  8.2
2.5 
n
Sample size, n = 41
Central Limit theorem
 Confidence interval (C.I.) for the population mean µ

 z  z 
C.I.   X   / 2 , X   / 2  1 0.95

 n n 
0.025
 /2
/2
0.025

-1.96
z / 2 z/2
1.96
 Example:
A random sample of size 100 is taken from a population with std. deviation
5.1, given that the sample mean is 21.6, construct a 95% confidence interval.

 1.96  5.1 1.96  5.1 

C.I.   21.6  , 21.6  
 100 100  Ans: (21.5, 22.6)
Distributions from Normal distribution
z1, z2, ……,zn Independent  2n  z12  z 22  ...  z 2n
standard normal random variables
P( X   2 , n )  

z  2 ,n
Standard normal distribution Chi-square distribution with
‘n’ degrees of freedom

n
x 1
1 x
 2
e   2
2 2
pdf  ,x  0
n 
  1!
2 
Distributions from Normal distribution
z – Random variable with standard  2 ,n – Random variable with Chi-square
normal distribution distribution

 

z
 2 ,n

P( t n  t  , n )  
z
tn  
 2n n
n
As n becomes large, n2  1  t n  z t  ,n
t- distribution with ‘n’ degrees of
freedom
Distributions from Normal distribution
For independent chi-square random
variables
 n and  m
2 2

,  2n
Fn ,m  n
 2 
m
m
0 F ,n ,m
P(Fn ,m  F ,n ,m )  
F- distribution with degrees
of freedom ‘n’ and ‘m’
1
 F1 ,m ,n
F ,n ,m
How to use these sampling distributions to
draw conclusion?
 Hypothesis testing
 Concerned with two distinct choices:
 Null Hypothesis (H0)
 Alternate hypothesis (H1)
 Test whether to accept or reject H0 using various test statistics.
 Two types of errors:

Two possibilities Decision

Accept H0 Reject H0
H0 True Correct ! Type I error
H1 True Type II error Correct !
Testing the hypothesis
 One tail or two tail?
Acceptance
region

Acceptance Rejection Rejection

1-α region Rejection
region region
region
1-α
α α/2 α/2
One tailed Two tailed

 Confidence level: 1-α : probability that the computed estimate

will lie in the acceptance region
 Level of significance: α :probability that the computed estimate
will lie in the rejection region
Distribution statistics in hypothesis testing
 Que.No.1: The spot speed at a particular location in an
expressway are known to be normally distributed with a mean
of 80km/hr. and std. dev. of 15km/hr. A new radar speed meter
was bought by traffic dept. and a set of 100 observations were
taken. The mean speed observed was 77.3km/hr. Is there any
evidence to prove that :
(i) the new speed meter might have been faulty
(ii) the new speed meter is showing lesser speed than actual.
Assume 5% level of significance.

Solution
Distribution statistics in hypothesis testing
 Que. No. 2: The mean spot speed of 15 vehicles observed on a
Sunday at a particular roadway was 81.2km/hr. The mean
speeds of all vehicles at this location as per previous records
was 75.5 km/hr. and std. dev. 10.2km/hr. Is there sufficient
evidence to show that the speeds of vehicles on that Sunday was
higher than the average speed? Take level of significance as
5%

Solution
Distribution statistics in hypothesis testing
 Ques. No.3: Two samples of speed data are collected are as
follows:
For sample 1, mean speed is 74.3km/hr. and std. dev. is 7km/hr. (n 1=120)
For sample 2, mean speed is 72.5km/hr. and std. dev. is 8km/hr. (n 2=120)
Is there any evidence to prove that the mean speed reduced by
more than 0.5km/hr. when using these samples? Assume level of
significance as 10%.

Solution
Distribution statistics in hypothesis testing
 Que.No.4:For a given vehicle speed data sample of size 20, the
standard deviation observed was 12.5km/hr. The data can be
used only if the standard deviation is near to approximately
equal to10km/hr. Check whether the data can be accepted at 5%
level of significance.

Solution
Distribution statistics in hypothesis testing
 Que.No.5:It is desired to determine whether there is less
variability in the speed data collected for day 1 than for day2.
If independent random samples are taken for these two days as
below:
For day 1: std. dev.=12km/hr. ;sample size=12
For day 2:std. dev.=10km/hr. ;sample size=14,
test the given hypothesis with a level of significance 5%.

Solution
Distribution statistics – Hypothesis testing
 Que.No.6: Every minute vehicle count data was collected for a period of 65
minutes. Determine at 95% confidence level , whether the data follows a
poisson distribution.
No. of Observed
arrival frequency
0 2
1 6
2 7 To test the fit of data to a
3 12 particular distribution,

4 13 ‘GOODNESS OF FIT’ test

5 9
6 9 Solution
7 4
8 2
9 1
Summary of test statistics for Hypothesis testing
TEST STATISTICS
H1
Reject H0 if
Hint: µ0 = population mean
ϭ0 = population std. dev.
Large sample – comparing mean
H0 :   0
  0 z  z 
X 
z   0 z  z

n   0 z  z  or z  z 
2 2

Small sample / unknown variance – comparing mean

H0 :   0
  0 t  t 
X 
t   0 t  t
s
n   0 t   t  or t  t 
2 2
Summary of test statistics for Hypothesis testing
TEST STATISTICS
H1
Reject H0 if
Hint: µ0 = population mean
ϭ0 = population std. dev.
Comparison of sample mean
H 0 : 1   2  

X z  z 
1  X 2    1   2 
1   2  
z
2
1  2
2
1   2   z  z

n1 n2 1   2   z  z  or z  z 
2 2

One variance
2
H 0 :  2  0
 2   2
2
 2  0

2 
 n  1 s 2
2  0
2
 2  12
2
 2  0
2
 2  12 or  2   2
2 2
Summary of test statistics for Hypothesis testing
TEST STATISTICS
H1
Reject H0 if
Hint: µ0 = population mean
ϭ0 = population std. dev.
Two variance
2 2
H 0 : 1   2

s1
2
/ n1 s 2
2
/n2  2
1   2
2
F  F ,n1 1,n 2 1
F
s 2
2
/ n  s
2 1
2
/n2 
2
1   2
2
F  F ,n 2 1,n1 1
s 2
l arg e / nL  s 2
small / n S  1   2
2 2
F  F ,n l arg e 1,n small 1
Underlying distribution
H 0 : Data follows given distribution

 
2  Oi  E i  2 Data does
not follow
i Ei given  2   2
distribution
Thank You