Submit Assignment For Help

Go To Answer Directly

ESD.86

info@statisticsassignmenthelper.com

Models, Data, Inference for Socio-Technical Systems

Massachusetts Institute of Technology

Problem
Engineering Set #3
Systems Division

Issued: Wednesday February 28, 2007. Due: Wednesday March 7, 2007 at 10:00am.

1. Max and Min. Consider two r.v.s X and Y that are uniformly distributed and
independent. Random variable X is uniform over the interval [0,1]. Y is uniform over the
interval [0,2].
Let
W=Min{X, Y}

Z =Max{X, Y}

(a) Find the joint pdf for W and Z.

(b) Find the pdf for W + Z.

2. Random Incidence. Building a Car. Imagine that you are working on an assembly
line and waiting for two components, fidgets and whoosies. As soon as you have the two
components you can complete the assembly of your reconstructed classic Pierce-Arrow
automobile1. These components arrive as independent random processes on adjacent
conveyor belts. Fidgets arrive as a homogeneous Poisson process with mean inter-arrival
time of 10 minutes. Whoosies arrive as a renewal process with time TW between
successive renewals found from this pdf:

fTW (t) = (1/2)(0.1)e"0.1t + (1/2)(0.02)e"0.02t t # 0.

The whoosie pdf is sometimes called the hyper-geometric pdf.

(a) Find the mean time between arrivals of both fidgets and whoosies.
(b) You!have been on lunch break and you arrive back at your workstation, with the
partially reconstructed Pierce-Arrow, and the fidget and whoosie arrival processes
in full swing.
a. Find the mean time from your arrival back at your station until the first
fidget arrives.
b. Find the mean time from your arrival back at your station until the first
whoosie arrives.
c. Find the mean time until both have arrived and you can complete your
reconstruction of the Pierce-Arrow.

1
For a brief history of the Pierce-Arrow car, see
http://www.antiquecar.com/index/listings/category766.htm

https://www.statisticsassignmenthelper.com/
1
3. Spatial Poisson on a Line. Ambulances patrol an infinitely long straight East-West
highway at spatial density of γ ambulances per mile. At any given time, the ambulances
are positioned along the highway as a homogeneous spatial Poisson process. Being
public safety vehicles, ambulances are allowed to make U-turns anywhere, so that their
direction of travel is not relevant. You have a traffic accident at point x along the
highway and with your new GPS device you summon two ambulances, the closest one to
the East of your location and the closest one to the West of your location. Once,
summoned, each ambulance drives from its location to your location at a constant speed
of 60 miles per hour to reach your location.
(a) Determine the pdf for the time until the first ambulance arrives. Calculate the
mean value.
(b) Determine the pdf for the time until the second ambulance arrives. Calculate the
mean value. Compare to the mean value found in part (a). Are you surprised in
any way? Comment.

4. Covered Rectangle. Homogeneous spatial Poisson processes have in two dimensions

the same property that homogeneous time Poisson processes have: Given that we have
conducted our probabilistic experiment and we know only the exact number of Poisson
entities within a fixed area, then – with no further information – the locations of these
entities are independent and uniformly distributed over the fixed area. This property is
analogous to the unordered arrival times in a homogeneous time-Poisson process, as
these unordered times are independent and uniformly distributed over the fixed time
interval of interest. Suppose that the fixed area of interest is a 2-mile-by-10-mile
rectangle, with sides parallel to the x and y axes, respectively. The 2-mile-by-10-mile
rectangle is situated within a larger community in which a homogeneous spatial Poisson
Process operates. Now suppose that the spatial Poisson process has distributed its entities
over the community. We are told only that the number of Poisson entities within the
rectangle is exactly 10. We define the coverage rectangle as the smallest rectangle
within the 2-by-10 mile rectangle that contains all 10 Poisson entities and that has sides
parallel to the larger rectangle. Find the expected area of the coverage rectangle.

5. Random Incidence – Cookies, etc. In your own ESD research or other professional
activity, think about the random incidence problems that are explained by the “chips in
chocolate chip cookies” setup. That is, in your work there is the potential for major
selection bias due to this phenomenon. Identify one focused topic in your research or
professional activity in which such selection bias is or could be an issue. Try to frame,
formulate and solve this problem is as complete detail as you can. Use the formulas
where appropriate. If you have data to support your work, even better! If you want to
display results on a spreadsheet, terrific! If no such threats occur in your professional
work, then think of your personal life and possibilities there for this to happen. If you
still draw a blank, then do the analysis for a situation you read about in the media.

2
 Problem Set #3

1. Max and Min – 4 pts

a) 2 pts
By observation, we know that W ≤ Z.

Two cases: on [0;1] and on [1;2].

On the interval [0;1]:

1 w2
FW ,Z (w,z) = [ zw + w(z " w)] = zw "
2 2
2
# FW ,Z (w,z)
fW ,Z (w,z) = =1
#w#z

On the interval [1;2]:

1 w + wz " w 2
! FW ,Z (w,z) = [ zw + w(1" w)] =
2 2
2
# FW ,Z (w,z) 1
fW ,Z (w,z) = =
#w#z 2

!
 b) 2 pts

By observation, we see that T = W + Z = X + Y. X and Y are independently distributed.

We have three cases: on [0;1], on [1;2] and on [2;3].

On the interval [0;1]:

1 t2 t2 t
FT = " = # fT =
2 2 4 2

On the interval [0;1]:

1 t + [1# (2 # t)] t 1 1
! FT = "1" = # $ fT =
2 2 2 4 2

!
 On the interval [0;1]:
2
1 $ (3 # t) 2 ' (3 # t ) * f = 3 # t
FT = "1" &2 # ) = 1# T
2 % 2 ( 4 2

2. Building a car – 5 pts

a) 2 pts
! Mean time between arrivals of fidgets:
1
E[TF ] = = 10 _ min
"
Mean time between arrivals of whoosies:
1&$ )
E[TW ] = ( % t " (0.1" e#0.1"t + 0.02 " e#0.02"t ) dt+ = 30 _ min
! 2' 0 *

b.1) 1 pt
The Poisson process being memoryless, the mean time from our return to the workstation until
! the first fidget arrives is 10 min.

b.2) 1 pt
Mean time to arrival of first whoosy:
# " W &2 2
E[TW ] ) E 2 [TW ] , 1 1 2/
1+ % ( 1+ 2 . 2 + 2 ) 30 1
$ E[TW ] ' E [TW ] + +2
E[VW ] = E 2 [TW ] = 30 2 = 15 * .1+ 1 1
2E[TW ] 60 . 30 2 1
.- 10
, 2500 + 100 ) 900 /
E[VW ] = 15 * .1+ 10 = 43.33_ min
- 900

b.3) 1pt
We can have two cases here: either the fidget arrives first, or it’s the whoosy.
! 43.3 43.3
P{Fidget _ first} = $ ".e # ".t
.dt = $ 0.1% e #0.1%t
= 0.987
0 0
If the fidget arrives first, then the total waiting time is the mean waiting time for the first whoosy

to arrive: 43.33 min.

If the whoosy is first, however, we will have to wait 10 more minutes on average, since the

! Poisson process is memoryless.

Therefore: E{Total _ waiting _ time} = 0.987 " 43.33 + (1# 0.987) " (43.33 + 10) = 43.46 _ min

!
 3. Spatial Poisson on a line – 4 pts
a) 2 pts
The distance to first ambulance obeys to a Poisson process with a pdf d1 (t) = " .e#" .t
D1 is the corresponding cumulative distribution function.
1
Therefore: E[D1 ] = _ mile is the average length we have to check before finding an ambulance.
"
!

E[D1]
!

Road
Accident

This distance is centered on the accident: the average distance from the accident to the nearest
ambulance (E[D’1]) is half this value. Since the ambulance is driven at 1 mile per minute,
1
E[T'1 ] = _ min
2."

b) 2 pts
1
! Similarly, we will again have to cover mile on average to find the next ambulance, starting on
"
the border of the area previously searched (memoryless property of the Poisson pdf). However,
since we have already determined on which side was the previous ambulance, we know that the
second ambulance is necessarily on the other direction.
!
1 1 3
Therefore, the distance to the second nearest ambulance is: E[D'2 ] = + = _ miles and
2." " 2."
1 1 3
E[T'2 ] = + = _ min
2." " 2."
!
4. Covered rectangle – 4 pts
! Coverage rectangle has sides of length L1 = [Max(X) – Min(X)] and L2 = [Max(Y) – Min(Y)].
Its area is A = L1 x L2.
10
$ x '10
FM (x) = P{Max(X) " x} = # P[X i " x] = & ) , x * [0;2]
i=1
% 2(
$ x '9
f M (x) = 5& )
% 2(
2
2
$ x '9 5 - x11 0 20
E[Max(X)] = , x + 5 + & ) dx =
%2(
/ 2 =
512 . 11 10 11
0

!
 10
& 2 # x )10
Fm (x) = P{Min(X) " x} = 1# P{Min(X) $ x} = 1# % P[X i $ x] = 1# ( + , x , [0;2]
i=1
' 2 *
& 2 # x )9
f M (x) = 5( +
' 2 *
2
& 2 # x )9 & 2 2
& 2 # x )10 )
2
5 (/ x(2 # x)10 2 5 /(2 # x)11 2 2
E[Min(X)] = . x - 5 -(
' 2 *
+ dx = (1 4 # .(
512 '0 #10 30 0 ' #10 *
+ dx ++ = 1 4 =
0 * 512 0 #110 30 11
20 2 18
Therefore, E[L1 ] = " =
11 11 11
10
$ y '10
! FM (y) = P{Max(Y ) " y} = # P[Yi " y] = & ) , y * [0;10]
i=1
% 10 (
! $ y '9
Similarly, f M (y) = & )
%2(
10
10
$ y '9 1 - y11 0 100
E[Max(Y )] = , y + & ) dy = 9 / 2 =
%10 ( 10 . 11 10 11
0
10
& 10 # y )10
Fm (y) = P{Min(Y ) " y} = 1# P{Min(Y ) $ y} = 1# % P[Yi $ y] = 1# ( + , y , [0;10]
i=1
' 2 *
! &10 # y ) 9
f M (x) = ( +
' 10 *
10
& 10 # y ) 9 & 10 10
&10 # y )10 ) 1 /(10 # y)11 2
10
1 (/ y(10 # y)10 2 10
E[Min(X)] = . y - ( + dy = 9 (1 4 # .( +
+ dy + = 9 1 4 =
0
' 10 * 10 '0 #10 30 0 ' #10 * * 10 0 110 30 11
100 10 90
Therefore,
E[L2 ] = " =
11 11 11
18 90 1620
! As a consequence,
E[A] = E[L1 ] " E[L2 ] = " = # 13.39 _ miles2 # 33.67 _ km 2
11 11 121
! 5. Cookies, etc – 3pts
No standardized answer for this question, obviously.
!