You are on page 1of 56

Continuous Random

Variables

Continuous Random
Variable
A

continuous random variable is one


for which the outcome can be any value
in an interval of the real number line.
Usually a measurement.
Examples
Let Y = length in mm
Let Y = time in seconds
Let Y = temperature in C
L. Wang, Department of Statistics
University of South Carolina; Slide 2

Continuous Random
Variable
We

dont calculate P(Y = y), we


calculate P(a < Y < b), where a and b
are real numbers.

For

a continuous random variable


P(Y = y) = 0.

L. Wang, Department of Statistics


University of South Carolina; Slide 3

Continuous Random
Variables
The probability density function (pdf)
when plotted against the possible values
of Y forms a curve. The area under an
interval of the curve is equal to the
probability that Y is in that interval.

0.40

f(y)

b Y

L. Wang, Department of Statistics


University of South Carolina; Slide 4

The entire area under a


probability density curve for a
continuous random variable
A.
B.
C.
D.

Is always greater than 1.


Is always less than 1.
Is always equal to 1.
Is undeterminable.
L. Wang, Department of Statistics
University of South Carolina; Slide 5

Properties of a Probability Density Function


(pdf)
1)
2)

f(y) > 0 for all possible intervals of y.

f ( y )dy 1

3)

If y0 is a specific value of interest, then


the cumulative distribution function (cdf)
is
y
F ( y0 ) P (Y y0 )

f ( y)dy

4)

If y1 and y2 are specific values of interest,


y
then P( y Y y ) f ( y )dy F ( y ) F ( y )
2

y1

L. Wang, Department of Statistics


University of South Carolina; Slide 6

Grams of lead per liter of


gasoline has the probability
density function: f(y) = 12.5y 1.25
for 0.1 < y < 0.5
What is the probability that the
next liter of gasoline has less
than 0.3 grams of lead?
L. Wang, Department of Statistics
University of South Carolina; Slide 7

Suppose a random variable Y


has the following probability
density function: f(y) = y
if
0<y<1
2-y if 1 < y<2
0
if 2 < y.
Find the complete form of the
cumulative distribution function
F(y) for any real value y.
L. Wang, Department of Statistics
University of South Carolina; Slide 8

Expected Value for a


Continuous Random Variable

Recall Expected Value for a discrete


random variable:

E (Y ) y p ( y )

Expected value for a continuous random


variable:

E (Y )

yf
(
y
)
dy

Department of Statistics
L.Wang,
University of South Carolina; Slide 9

Variance for Continuous


Random Variable
Recall: Variance for a discrete random
variable:

Var (Y ) ( y ) p( y )
2

Variance for a continuous random


variable:

Var (Y ) ( y ) f ( y )dy
2

L. Wang, Department of Statistics

University of South Carolina; Slide 10

Difference between Discrete


and continuous random
variables
Possible values that can be
assumed
Probability distribution function
Cumulative distribution function
Expected value
Variance
L. Wang, Department of Statistics
University of South Carolina; Slide 11

Times Between Industrial


Accidents
The

times between accidents for a


10-year period at a DuPont facility
can be modeled by the exponential
distribution.

f ( y ) e

y 0 and 0

where is the accident rate (the


expected number of accidents per day
in this case)
L. Wang, Department of Statistics
University of South Carolina; Slide 12

Example of time between


accidents

Let Y = the number of days between two


accidents.
Time
12 days

Accident
#1 #2

35 days

Accident
#3

5 days

Accident

L. Wang, Department of Statistics


University of South Carolina; Slide 13

Times Between Industrial


Accidents

Suppose

in a 1000 day period there


were 50 accidents.

= 50/1000 = 0.05 accidents per


day

or

1/ = 1000/50 = 20 days between


accidents
L. Wang, Department of Statistics
University of South Carolina; Slide 14

What is the probability that this


facility will go less than 10 days
between the next two accidents?

f(y) = 0.05e-0.05y

L. Wang, Department of Statistics


University of South Carolina; Slide 15

10

P (Y 10) F (10) 0.05e 0.05 y dy


0

Recall:

e du e
u

F (10) e

0.05 y 10
0

| 0.39
L. Wang, Department of Statistics
University of South Carolina; Slide 16

In General
y

P (Y y ) e dt
t

P (Y y ) F ( y ) e

t y
0

| 1 e

P (Y y ) 1 F ( y ) e

L. Wang, Department of Statistics


University of South Carolina; Slide 17

Exponential Distribution

1 e

L. Wang, Department of Statistics


University of South Carolina; Slide 18

If the time to failure for an electrical


component follows an exponential
distribution with a mean time to failure
of 1000 hours, what is the probability
that a randomly chosen component
will fail before 750 hours?
Hint: is the failure
rate (expected number
of failures per hour).
L. Wang, Department of Statistics
University of South Carolina; Slide 19

Mean and Variance for an


Exponential Random Variable

E (Y ) ye

Var (Y ) y e
2

1
dy

1
dy

1
2

Note: Mean = Standard


L. Wang, Department of Statistics
Deviation

University of South Carolina; Slide 20

The time between accidents at a


factory follows an exponential
distribution with a historical
average of 1 accident every 900
days. What is the probability that
that there will be more than 1200
days between the next two
accidents?

L. Wang, Department of Statistics


University of South Carolina; Slide 21

If the time between accidents


follows an exponential distribution
with a mean of 900 days, what is
the probability that there will be
less than 900 days between the
next two accidents?

L. Wang, Department of Statistics


University of South Carolina; Slide 22

Relationship between
Exponential & Poisson
Distributions
Recall

that the Poisson distribution is


used to compute the probability of a
specific number of events occurring
in a particular interval of time or
space.
Instead of the number of events
being the random variable, consider
the time or space between events as
the random variable.
L. Wang, Department of Statistics

University of South Carolina; Slide 23

Relationship between
Exponential & Poisson
Exponential distribution models time
(or space) between Poisson events.

TIME
L. Wang, Department of Statistics
University of South Carolina; Slide 24

Exponential or Poisson
Distribution?

We model the number of industrial


accidents occurring in one year.

We model the length of time between two


industrial accidents (assuming an accident
occurring is a Poisson event).

We model the time between radioactive


particles passing by a counter (assuming a
particle passing by is a Poisson event).

We model the number of radioactive


particles passing by a counter in one hour

L. Wang, Department of Statistics


University of South Carolina; Slide 25

Recall: For a Poisson


Distribution

( t ) y e t
P(Y y ) p ( y )
y!

y = 0,1,2,

where is the mean number of events per


base unit of time or space and t is the
number of base units inspected.
The probability that no events occur in a
span of time (or space) is:

( t ) y e t ( t ) 0 e t
t
p ( 0)

e
y!
0!
L. Wang, Department of Statistics
University of South Carolina; Slide 26

Now let T = the time (or space)


until the next Poisson event.

P(T t ) e

In other words, the probability that


the length of time (or space) until
the next event is greater than some
given time (or space), t, is the
same as the probability that no
events will occur in time (or space)
L. Wang, Department of Statistics
t.
University of South Carolina; Slide 27

Radioactive Particles
The

arrival of radioactive particles at a


counter are Poisson events. So the
number of particles in an interval of
time follows a Poisson distribution.
Suppose we average 2 particles per
millisecond.
What is the probability that no particles
will pass the counter in the next 3
milliseconds?
What is the probability that more than
3 millisecond will elapse before the
next particle passes?L. Wang, Department of Statistics

University of South Carolina; Slide 28

Machine Failures
If the number of machine failures in a
given interval of time follows a
Poisson distribution with an average of
1 failure per 1000 hours, what is the
probability that there will be no
failures during the next 2000 hours?
What is the probability that the time
until the next failure is more than
2000 hours?

L. Wang, Department of Statistics


University of South Carolina; Slide 29

Number

of failures in an interval of time


follows a Poisson distribution. If the mean
time to failure is 1000 hours, what is the
probability that more than 2500 hours
will pass before the next failure occurs?

A. e-4
B. 1 e-4
C. e-2.5
D. 1 e-2.5
L. Wang, Department of Statistics
University of South Carolina; Slide 30

Challenging
questions

If ten of these components are used in


different devices that run
independently, what is the probability
that at least one will still be operating
at 2500 hours?
What about he probability that exact 3
of them will be still operating after
2500 hours?
L. Wang, Department of Statistics
University of South Carolina; Slide 31

Normal Distribution

f(y)

f(y) =
E[Y] =

1
( y ) 2 / 2 2
e
, y
2
and

2
Var[Y]
=

L. Wang, Department of Statistics


University of South Carolina; Slide 32

Normal Distribution

Characteristics
Bell-shaped curve
- < y < +
determines distribution location
and is the highest point on curve
Curve is symmetric about
determines distribution spread
Curve has its points of inflection at
+
L. Wang, Department of Statistics
University of South Carolina; Slide 33

Normal Distribution

-4

-3

-2

-1

L. Wang, Department of Statistics


University of South Carolina; Slide 34

Normal Distribution
N( = 5, = 1)

N( = 0, = 1)
f(y)

-4

-3

-2

-1

L. Wang, Department of Statistics


University of South Carolina; Slide 35

Normal Distribution
N( = 0, = 0.5)
f(y)

N( = 0, = 1)

-4

-3

-2

-1

y L. Wang, Department of Statistics


University of South Carolina; Slide 36

Normal Distribution
N( = 5, = 0.5)
N( = 0, = 1)
f(y)

-4

-3

-2

-1

L. Wang, Department of Statistics


University of South Carolina; Slide 37

68-95-99.7 Rule
0.997
0.95
0.68

-4

-3
-3

-2
-2

+ 1 covers
approximately 68%

-1
-1

1
2
+1
+2

+ 2 covers
approximately 95%

3
+3

+ 3 covers
approximately99.7%

L. Wang, Department of Statistics


University of South Carolina; Slide 38

Earthquakes in a California
Town

Since 1900, the magnitude of


earthquakes that measure 0.1 or
higher on the Richter Scale in a
certain location in California is
distributed approximately normally,
with = 6.2 and = 0.5, according
to data obtained from the United
States Geological Survey.
L. Wang, Department of Statistics
University of South Carolina; Slide 39

Earthquake Richter Scale


Readings

34%

34%

2.5%
13.5%
-4

-3

57

-2
5.2

-1
5.7

13.5%
0
6.2

1
6.7

68%

2
7.2

2.5%
3

159

95%
L. Wang, Department of Statistics
University of South Carolina; Slide 40

Approximately what percent of the


earthquakes are above 5.7 on the Richter
Scale?

34%

34%

2.5%

2.5%
13.5%

-4

-3

-2
5.2

-1
5.7

13.5%
0
6.2

1
6.7

2
7.2

68%
95%
L. Wang, Department of Statistics
University of South Carolina; Slide 41

The highest an earthquake can


read and still be in the lowest
2.5% is _.
34%

34%

2.5%
13.5%
-4

-3

-2
5.2

-1
5.7

13.5%
0
6.2

1
6.7

2
7.2

2.5%
3

68%
95%
L. Wang, Department of Statistics
University of South Carolina; Slide 42

The approximate probability an


earthquake is above 6.7 is
______.
34%

34%

2.5%
13.5%
-4

-3

-2
5.2

-1
5.7

13.5%
0
6.2

1
6.7

2
7.2

2.5%
3

68%
95%
L. Wang, Department of Statistics
University of South Carolina; Slide 43

Standard Normal Distribution


Standard

normal distribution is the


normal distribution that has a mean
of 0 and standard deviation of 1.
N( = 0, =
1)

-4

-3

-2

-1

L. Wang, Department of Statistics


University of South Carolina; Slide 44

Z is Traditionally used as the


Symbol for a Standard Normal
Random Variable

-4

-3

4.7

-2

-1

5.2

5.7

6.2

6.7

7.2

7.7

L. Wang, Department of Statistics


University of South Carolina; Slide 45

Normal Standard Normal


Any normally distributed random
variable can be converted to standard
normal using the following formula:

y
Z

We can compare observations from two


different normal distributions by
converting the observations to standard
normal and comparing the
L. Wang, Department of Statistics
standardized observations.
University of South Carolina; Slide 46

What is the standard normal


value (or Z value) for a
Richter reading of 6.5?
Recall Y ~ N(=6.2, =0.5)

L. Wang, Department of Statistics


University of South Carolina; Slide 47

Example

Consider two towns in California. The


distributions of the Richter readings over 0.1
in the two towns are:
Town 1:
Town 2:

X ~ N( = 6.2, = 0.5)
Y ~ N( = 6.2, = 1).

- What is the probability that Town 1 has an


earthquake over 7 (on the Richter scale)?
- What is the probability that Town 2 has an
earthquake over 7?
L. Wang, Department of Statistics
University of South Carolina; Slide 48

Town 1

Town 2

0.21
2

0.05
5
Z
X

-4

-3

4.7

-2

5.2

-1

5.7

6.2

6.7

7.2

7.7

Z
Y

-4

-3

-2

3.2 4.2

-1

5.2

6.2

7.2 8.2

9.2

7 6 .2
P ( X 7 .0 ) P Z
P ( Z 1.6) 0.055
Town 1:
0 .5

7 6.2
Town 2: P (Y 7.0) P Z 1.0 P ( Z 0.8) 0.212

Department of Statistics
L. Wang,

University of South Carolina; Slide 49

Standard Normal

0.10

0.10

0.05

0.05

0.025
0.01
0.005
-4

-3

0.025
0.01
0.005
-2

-1

-2.326 -1.645

-2.576 -1.96 -1.282

1.645 2.326
1.282 1.96 2.576

L. Wang, Department of Statistics


University of South Carolina; Slide 50

The

thickness of a certain steel bolt


that continuously feeds a
manufacturing process is normally
distributed with a mean of 10.0 mm
and standard deviation of 0.3 mm.
Manufacturing becomes concerned
about the process if the bolts get
thicker than 10.5 mm or thinner than
9.5 mm.
Find the probability that the thickness
of a randomly selected bolt is > 10.5
L. Wang, Department of Statistics
or < 9.5 mm.

University of South Carolina; Slide 51

Inverse Normal Probabilities

Sometimes we want to answer a question


which is the reverse situation. Here we
know the probability, and want to find the
corresponding value of Y.

Area=0.0
25

y=?

L. Wang, Department of Statistics


University of South Carolina; Slide 52

Inverse Normal Probabilities

Approximately 2.5% of the bolts produced


will have thicknesses less than ______.

0.02
5
Z
Y

-4

-3

-2

-1

L. Wang, Department of Statistics


University of South Carolina; Slide 53

Inverse Normal Probabilities

Approximately 2.5% of the bolts produced


will have thicknesses less than ______.

Y 10.0
2
y 9.4
0.3
L. Wang, Department of Statistics
University of South Carolina; Slide 54

Inverse Normal Probabilities

Approximately 1% of the bolts produced


will have thicknesses less than ______.

0.01

Z
Y

-4

-3

-2

-1

L. Wang, Department of Statistics


University of South Carolina; Slide 55

Inverse Normal Probabilities

Approximately 1% of the bolts produced


will have thicknesses less than ______.

Y 10.0
2.326
y 9.3
0.3

L. Wang, Department of Statistics


University of South Carolina; Slide 56