Random Var Dist
Random Var Dist
Random variables, or rv’s are used mainly for the purpose to quantify events. So that probability
distribution will be with respect to a number, or a scale. The resulted study is more analytic.
Sometimes the events are themselves numerical in nature. Number of objects such as vehicles, incidences.
Measurements such as weights or heights. We will use expressions in the rv’s to replace events.
Example. A special tire are made in pairs due to the machine. Now let the rv X be the number of non-
defective tires in a pair. Then we have:
Expression Event
X=0 None is non-defective.
X=1 Exactly one is non-defective.
X=2 Exactly two are non-defective.
X≤1 No more than one is non-defective.
X<2 Less than two…
X≥1 No less than one ….
Example. Let the rv be the monthly precipitation in a certain region. Or the average temperature of a
region. It is straight-forward to see the event X > 30, or 10 <X < 25. With suitably defined pdf,
probabilities, that is, areas, can be calculated with integration.
There are two types of random variables, discrete and continuous. The differences between them are self-
explained. One must be careful since some of the popular quantities seem to be discrete but actually are
continuous. Being seemingly discrete is the result from rounding.
Still, the aim is at the distribution, probability distribution, now.
It is senseless to ask what value a random variable will turn out to be. Because it’s uncertain. Rather, the
target is the probability distribution.
The type of distributions for the two kinds of r.v.’s are different. We use tables for the discrete and curves
for the continuous.
It is really not that simple.
Discrete rv’s take on separated numbers. They can be either the number or not at all. So for the discrete,
we use probability “mass” at the possible values to show the likelihood.
Example 1
x P(X=x)
0 0.2 One can plot the
1 0.3 distribution.
0 1 2
2 0.5
Example 2 A branch office of a delivery company recorded the number of packages brought in by individual
customers to formulate the following distribution table.
x 1 2 3
P(X = x) 0.5 0.3 0.2
This kind of distribution is called the probability mass functions, pmf, or simply pf.
Continuous rv’s can be any number in an interval. The probability can no longer be evaluated with respect
to points.
So, instead, the probability will be expressed with areas, areas under a curve representing the distribution.
b
Example 3:
x P(a ≤ x ≤ b)= a
∫
f ( x)dx
P ( X≤x ) =∫−∞ f (s)ds
P(X=x) f(x)
a b
The function f(x) is the probability density function, pdf. It is equivalent to the ordinary distribution curve
we used to look at. The area under the curve on an interval [a, b] is defined to be the probability P(a ≤X≤b).
Quite different from the idea for discrete rv’s. Since any point doesn’t occupy area, it doesn’t matter
whether < or ≤ is used here.
For any continuous type r.v. X, P(X=x)=0, ∀x.
Conditions for probability mass functions.
Conditions for probability density functions.
∞
∫
(1) f(x) ≥ 0 (2) −∞
f (x)dx=1
(CDF) Sometimes it is easier to use the CDF of a rv to figure out the probabilities.
Cumulative dist. function CDF is defined to be
F(x)=P(X ≤ x).
When continuous, P(a ≤ x ≤ b) = P(a < x ≤ b)=…= etc. Once f(x) is known, we know that P(a ≤ x ≤ b)=
b x
∫a f ( x)dx , and the CDF, F(x)= ∫−∞ f ( s)ds . Then we have P(a ≤ x ≤ b)=F(b)−F(a). (This is actually the
Fundamental Theorem of Calculus.)
Conditions for a CDF.
(1) Non-deceasing. (2) F(−∞ ¿=0 , F (∞ )=1.
Points about cumulative distribution function (CDF) of random variables.
Cumulative distribution functions (CDF’s) seem to be an even better tool in calculating probabilities.
Beginners tend to accept probability functions better than they do CDF’s since the former gave better
representation of the idea of distribution. Indeed, there are something about CDF’s which make them
repulsive. Here they are.
Example1. CDF of a discrete random variable with the following probability function.
x 0 1 2 3 4 5
P(X=x) 0.18 0.39 0.24 0.11 0.04 0.04
P(X≤ x ¿ 0.18 0.57 0.81 0.92 0.96 1.00
Example2. The CDF of a Uniform distribution over [a, b] is defined as
{
0, if x < a,
x-a
F(x)= , if a ≤ x < b,
b-a
1, if b ≤ x.
{
0, if x<0,
8x
F(x)= , if 0≤x<30,
30+7x
1, if 30≤x.
Example4. This is EXP(0.04). The CDF is
F(x)=1- e -0.04x .
If, say, P(X<20) is desired, draw a vertical line at x=20. The y-coordinate of the intersection point with the
CDF is your answer.
P(X<20)=0.55067, P(X<60)=0.90928, P(20<X<60)=0.90928-0.55067=0.35861
If a certain percentile is desired, do it in the reverse order.
Note that for discrete CDF, it often happens that horizontal lines will not cross the CDF. It goes back to the
definition that the rth percentile is the lower bound of all the quantities such that r% of the elements fall
below.
You see now why percentiles were not discussed in details for discrete rv’s.
Take a look at the standard normal distribution, the CDF is like this.
Theoretically, it takes integration to calculate the areas under curves. Thanks to computer systems, we
don’t have to struggle with the calculation. With a well-prepared CDF table, or a simple CDF itself, we are
able to calculate the probabilities desired.
介紹計算方法(不用積分的情況下)
Example Exponential distribution with λ .
P( X ≤a )=1−e−λa
P ( a≤ X ≤b )=e− λb− e− λa
Suppose λ=2.5 , then P(X<2)=_____________.
Suppose λ=2.5 , the median of the distribution is ______________.
Example Suppose the shelf-life of a certain item, in days, has the following CDF.
{
0 ,if x ,
F ( x )= , if 0< x <30 ,
1 ,if x .
Bivariate distribution.
Joint distribution of two rv’s.
(Joint distribution of two continuous rv’s is not intensively discussed here because it’s not for elementary
Statistics.)
Example: Glass bottles. X is the number of bubbles. Y is the number of foreign particles.製
Y
0 1 X P(X=x) Y P(Y=y)
0 0.60 0.10 0.70 0 0.7 0 0.8.
X 1 0.12 0.08 0.20 1 0.2 1 0.20
2 0.08 0.02 0.10 2 0.1
0.80 0.20 1.00
Marginal distributions.
Joint distribution.
Parameters of the distribution. Mean and variance.
Expected value of a function of rv’s.
Calculation of parameters of probability distributions starts here.
E[X]= ∑ xP ( X=x )
σ 2 {X }= Ε[( x−Ε[ X ])2 ]=∑ ( x−Ε [ X ] )2 P( X =x )=Ε [ X 2 ]−( Ε [ X ] )2
2 2
(What about { Χ̄ , S ¿ ? { Χ̄ , S ¿ are sample statistics. Think about how they are calculated.)
Different symbols popularly used.
= E[X] = μ x
σ 2=σ 2 {X }=Ε[ ( x−Ε [ X ] )2 ]=Ε [ X 2 ] −( Ε [ X ] )2 =Var ( X )=σ 2x
When the random variable is of the continuous type, integration is used instead of summation. The formula
changed slightly to the following.
∞ ∞ 2
∫
E[X]= −∞
2 2 2
∫
xf (x)dx , σ {X }= Ε[( x−Ε[ X ]) ]=Ε [ X ]−( Ε [ X ] ) , where E[ X 2 ]= −∞ x f (x)dx .
2
Joint dist.
Example:
Y Y
0 1 0 1
X 0 0.60 0.10 0.70 X 0 x=0∩y=0 x=0∩y=1 x=0
1 0.12 0.08 0.20 1 x=1∩y=0 x=1∩y=1 x=1
2 0.08 0.02 0.10 2 x=2∩y=0 x=2∩y=1 x=2
0.80 0.20 y=0 y=1
P(x=1∩y=1)= P(x=0∩y=1)=
P(x=2)= P(x=0|y=0)=
P(y=0)= P(y=1|x≤1)=
Conditional distribution of X when Y=0, say.
Conditional distribution of Y when X=1, say.
The definition of expectation extends naturally to E[g(X,Y)]=∑ g(x, y)P(X = x ∩ Y = y). Note that the joint
probabilities are used for the weight here. If, in case g(x, y)=g 1 ( x ) + g2 (y), then the computation can be
simplified to using marginal probability distributions.
Review:If, and only if, P(X=i ∩ Y=j) =P(X=i)P(Y=j) ∀i, j, then we say that X and Y are independent.
(Independent is a strong property between two rv’s.)
Obviously the following example is not about two independent rv’s. Let’s look into the relationship revealed
by the joint distribution.
Example: X P(X=x) xP(X=x) x2 P(X=x)
0 0.70 0 0
1 0.20 0.20 0.20 E[X]=0.40
2 0.10 0.20 0.40 σ 2 {X } =0.60-0 . 402
E[X]=0.40 2 =0.44
0.60= E[ x ]
Example:
Y Y Y
0 1 0 21 2 0 1 2
X 0 0.0 0.05 0.25
0 0.1 0. 0.1 0 0.2 0.05 0
1 5
1 0.0 0.30 0.05 1 0.1 0. 0.1 1 0.0 0.30 0.05
5 2 5
2 0.2 0.05 0.0 2 0.1 0. 0.1 2 0 0.05 0.25
5 1
E[X]= E[Y]= {X,Y}= ________ ________ ________
{X}= {Y}= ρ XY = ________ ________ ________
(1)1-3 The shelf life of 8GB memory disk of a certain brand, in days, is considered a continuous random
variable X. The cumulative distribution function (CDF) F(x) is defined as follows.
{
1 for x>30,
8x
F(x)= for 0≤x≤30,
30+7x
0 for x<0.
(2)1-3 The shelf life of 8GB memory disk of a certain brand, in days, is considered a continuous random
variable X. The cumulative distribution function (CDF) F(x) is defined as follows.
(3)1-3 The shelf life of 8GB memory disk of a certain brand, in days, is considered a continuous random
variable X. The cumulative distribution function (CDF) F(x) is defined as follows.
{
1 for x>28,
x-12
F(x)= for 12≤x≤28,
28-12
0 for x<12.
1(26%) Warren Inc., one of the major manufacturers of TV sets in the country, is making production plans for
the coming season. Let the random variables X and Y denote forecast sales in million dollars in the coming
season for TV set models A and B, respectively. The joint probability distribution table is as follows.
Y
X 1 2 3 4
1 0.080 0.070 0.070 0.060
2 0.075 0.070 0.050 0.075
3 0.050 0.060 0.070 0.080
4 0.035 0.050 0.080 0.025
a(4%) What is P(X2Y2)?
b(12%) Compute the means and standard deviations of X and Y.
c(10%) Compute the covariance and the correlation coefficient of X and Y.
2(12%)The joint probability distribution table of random variables X and Y is given as follows.
Y
X 1 2 3
1 0.27 0.21 0.05
2 0.14 0.12 0.04
3 0.08 0.07 0.02
The following is also given.
Random variable Mean Variance
X 1.64 0.5704
Y 1.62 0.4556
a(6%) Compute the covariance {X, Y} of random variables X and Y. (2.68-1.64x1.62=0.0232)
b(6%) Compute the correlation coefficient {X, Y} of random variables X and Y. (0.0455)
Formulas:
Def. E[X+a]=E[X]+a
σ { X + a }=σ { X }
2 2
Def. E[cX]=cE[X]
σ { cX }=c σ { X }
2 2 2
σ { c1 X 1 , c 2 X 2 } =c 1 c 2 σ { X 1 , X 2 }
Def. E[X±Y]=E[X]±E[Y]
σ { X ± Y }=σ { X }+ σ { Y } ± 2 σ { X ,Y }
2 2 2
Families of probability distributions. They come in two groups, discrete and continuous.
(1) Discrete type
Discrete Uniform Distribution
Bernoulli(p)
It is denoted by X ~ Bernoulli(p), 0 and 1 are the possible values for X. The distribution is used to model
any trial with only two possible outcomes. The outcomes are denoted by Success (S) and Failure (F).
The rv X is the number of S, thus X=1 means getting a Success, and X=0 a Failure. The distribution
table, or function, goes
X P(X=x)
0 1-p E[X]= p and σ²{X}=p(1-p) can be computed.
1 p
When n identical Bernoulli trials are independently repeated, the resulted rv,s are called a Bernoulli
process, and the sum of them gives rise to a popular distribution, the Binomial distribution.
Binomial (n, p)
The sum of a Bernoulli process has a Binomial (n, p) distribution where n is the number of repetitions, and
the p is the overall success rate. The sum X is the total number of successes out of n trials. The
probability function is
Following is a part of Bin (n, p) probability function table. Notice the special correspondence of
probabilities between p and 1-p.
n p 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 x n
6 0 .5314 .3771 .2621 .1780 .1176 .0754 .0467 .0277 .0156 6
1 .3543 .3993 .3932 .3560 .3025 .2437 .1866 .1359 .0938 5
2 .0984 .1762 .2458 .2966 .3241 .3280 .3110 .2780 .2344 4
3 .0146 .0415 .0819 .1318 .1852 .2355 .2765 .3032 .3125 3
4 .0012 .0055 .0154 .0330 .0595 .0951 .1382 .1861 .2344 2
5 .0001 .0004 .0015 .0044 .0102 .0205 .0369 .0609 .0938 1
6 .0001 .0002 .0007 .0018 .0041 .0083 .0156 0 6
7 0 .4783 .3206 .2097 .1335 .0824 .0490 .0280 .0152 .0078 7
1 .3720 .3960 .3670 .3115 .2471 .1848 .1306 .0872 .0547 6
2 .1240 .2097 .2753 .3115 .3177 .2985 .2613 .2140 .1641 5
3 .0230 .0617 .1147 .1730 .2269 .2679 .2903 .2918 .2734 4
4 .0026 .0109 .0287 .0577 .0972 .1442 .1935 .2388 .2734 3
5 .0002 .0012 .0043 .0115 .0250 .0466 .0774 .1172 .1641 2
6 .0001 .0004 .0013 .0036 .0084 .0172 .0320 .0547 1
7 .0001 .0002 .0006 .0016 .0037 .0078 0 7
8 0 .4305 .2725 .1678 .1001 .0576 .0319 .0168 .0084 .0039 8
1 .3826 .3847 .3355 .2670 .1977 .1373 .0896 .0548 .0312 7
2 .1488 .2376 .2936 .3115 .2965 .2587 .2090 .1569 .1094 6
3 .0331 .0839 .1468 .2076 .2541 .2786 .2787 .2568 .2188 5
4 .0046 .0185 .0459 .0865 .1361 .1875 .2322 .2627 .2734 4
5 .0004 .0026 .0092 .0231 .0467 .0808 .1239 .1719 .2188 3
6 .0002 .0011 .0038 .0100 .0217 .0413 .0703 .1094 2
7 .0001 .0004 .0012 .0033 .0079 .0164 .0312 1
8 .0001 .0002 .0007 .0017 .0039 0 8
n 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 p
二項分配 Binomial(n, p)的累積機率 P(X≤ x ¿(CDF cannot be flipped as the pmf)
n p 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
6 0 0.5314 0.3771 0.2621 0.178 0.1176 0.0754 0.0467 0.0277 0.0156
1 0.8857 0.7764 0.6553 0 0.4201 0.3191 0.2333 0.1636 0.1094
2 0.9841 0.9526 0.9011 0.534 0.7442 0.6471 0.5443 0.4416 0.3438
3 0.9987 0.9941 0.9830 0 0.9294 0.8826 0.8208 0.7448 0.6563
4 0.9999 0.9996 0.9984 0.830 0.9889 0.9777 0.9590 0.9309 0.8907
5 1.0000 1.0000 0.9999 6 0.9991 0.9982 0.9959 0.9918 0.9845
6 1.0000 0.962 0.9998 1.0000 1.0000 1.0001 1.0001
4
0.995
4
0.999
8
1.000
0
7 0 0.4783 0.3206 0.2097 0.133 0.0824 0.0490 0.0280 0.0152 0.0078
1 0.8503 0.7166 0.5767 5 0.3295 0.2338 0.1586 0.1024 0.0625
2 0.9743 0.9263 0.8520 0.445 0.6472 0.5323 0.4199 0.3164 0.2266
3 0.9973 0.9880 0.9667 0 0.8741 0.8002 0.7102 0.6082 0.5000
4 0.9999 0.9989 0.9954 0.756 0.9713 0.9444 0.9037 0.8470 0.7734
5 1.0001 1.0001 0.9997 5 0.9963 0.9910 0.9811 0.9642 0.9375
6 1.0002 1.0001 0.929 0.9999 0.9994 0.9983 0.9962 0.9922
7 5 1.0001 1.0000 0.9999 0.9999 1.0000
0.987
2
0.998
7
1.000
0
1.000
1
8 0 0.4305 0.2725 0.1678 0.100 0.0576 0.0319 0.0168 0.0084 0.0039
1 0.8131 0.6572 0.5033 1 0.2553 0.1692 0.1064 0.0632 0.0351
2 0.9619 0.8948 0.7969 0.367 0.5518 0.4279 0.3154 0.2201 0.1445
3 0.9950 0.9787 0.9437 1 0.8059 0.7065 0.5941 0.4769 0.3633
4 0.9996 0.9972 0.9896 0.678 0.9420 0.8940 0.8263 0.7396 0.6367
5 1.0000 0.9998 0.9988 6 0.9887 0.9748 0.9502 0.9115 0.8555
6 1.0000 0.9999 0.886 0.9987 0.9965 0.9915 0.9818 0.9649
7 1.0000 2 0.9999 0.9998 0.9994 0.9982 0.9961
8 0.972 1.0000 1.0000 1.0001 0.9999 1.0000
7
0.995
8
0.999
6
1.000
0
n 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50
If Y~ Binomial(n, p), then E[Y]=np and σ²{Y}=np(1-p).
( An easy way to show it is that Y= x 1 , x 2 ,…, x n where E[ x i ]= p and σ²{ x i }=p(1-p).
x
While i ’s are indep., we thus have E[Y]=np and σ²{Y}=np(1-p).)
Let’s try some calculation.
a. X~Bin(20,0.2) P(X=4)=
b. X~Bin(20,0.2) P(X≥4)=
c. X~Bin(20,0.7) P(X=15)=
d. X~Bin(20,0.7) P(X≥15)=
This is good time to let you know how useful CDF’s really are.
Hypergeometric dist.
P (Y = y )=
( y n− y )
NS
)( N F
( Nn )
The distribution involves picking n items from a collections of N items in which
N S are successes and
N F are Failures. That is N S + N F =N. The rv Y is for the number of Successes out of the n picked. The
NS NS NS
N −n
parameters are E[Y]=n N and σ²{Y}=n N (1- N ) N −1 .
Geometric dist.
Suppose we stop as soon as we pick a Success during a Bernoulli process. The number of picks Y is said to
1
1− p
y−1
have a Geometric distribution. The probability function is P(Y=y)= p(1− p) ,y=1,2,…,E[Y]= p ,σ²{Y}= p2
. Note that p is ________________.
It is good practice trying to calculate the parameters of the distribution.
2 3
3 4
4 5
5 6
6 7
7 8
8 9
(
P(Y=y)= k −1
)
y −1 p (1−p )
k y −k
,y= k, k+1,…
k( 1− p )
k
2
The parameters are E[Y]= and σ²{Y}= p
p
Poisson distribution
(Poisson, 1781 ~ 1840, Galois, 1811 ~ 1832)
The rv X is usually used to denote number of occurrences of a certain event during a given range which can
μx
e−μ
either be in space or in time. If X~Poisson(µ), then P(X=x)= x ! , x= 0,1,2,…. The mean is E[X]= µ and
the variance is σ²{X}=µ. Note that µ is the mean occurrence counts. It can be interpreted in a number of
ways.
We should not be concerned that there is no upper limit to the rv Y since the probability will vanish fairly
reasonably quick.
Let’s compute the probability functions for μ=5 .
x P(X=x)
0
1
One special feature: The sum of two independent Poisson rv’s will be again Poisson with a mean the sum of
the means of the component distributions. That is, if independent,
Continuous Distributions
In continuous case, probability is evaluated in intervals as areas. There is no probability mass at any single
point, that is P(X=x)=0 for all x. So from now on, P(X< x)=P(X≤x), that is, “less than” is the same as “less than
or equal to” for all similar cases.
The distribution is represented by a function, the pdf, and the area under the function curve on a specified
interval [a, b] is evaluated for the corresponding probability P(a≤X≤b).
The required condition for pdf’s.
Condition 1: ______________________________
Condition 2: ____________________________________
E[X]= σ 2 {X } =
m
Let X~ N(, ²) in any general case, finding P(a≤X≤b) is one work. We transform any given normal rv
x −μ
with standardization to get the Z-score, that is, Z= σ , and Z~N(0,1) according to a theorem. We
calculated and prepare the the CDF of N(0, 1) which is enough for us to deal with all normal cases.
Say, we have X~N(55, 20), what is P(48≤X≤63)=?
48−55 63−55
P(48≤X≤63)= P( √20 ≤X≤ √ 20 )=P(_____< Z < ______)=_________
Standardization will enable us to find probabilities regarding any, any normal random variable at all.
Calculate probabilities.
Let Z~N(0,1), find the following probabilities.
a. P(Z≤1.62)= b. P(Z≥1.03)=
c. P(Z≤-0.89)= d. P(Z≥-2.02)=
e. P(-0.1 ≤Z ≤1.22)= f. P(0.03 ≤Z ≤1.83)
Calculate percentiles. Let Z(a) be the ath percentile of N(0, 1), that is, P(Z ≤Z(a))=a.
Let Z~N(0,1), find the following percentiles.
a. Z(0.7580)= b. Z(0.2420)=
c. Z(0.9750)= c. Z(0.0250)=
Example 2: The time between arrivals of a bus at the bus stop is exponential with an arrival rate of 2.5 per
half an hour. Sam is waiting for the bus now. What is the probability that Sam gets on to a bus
within 30 minutes?
Example 3: What is the Q3 of Exp(0.25)? Exp(0.1)? Exp(1.2)?
2 The flow of cars entering Tunghai University through the IBA entrance during weekdays can be thought of
as a Poisson process with the intensity of =20 cars per hour. Let X be the number of cars that enter the
IBA entrance during the period from 2:30pm to 3:50pm this coming Thursday. What is the probability
P(22 X < 26)? .2646
3 Assume that the random variable X has an exponential distribution. It is also known that the third quartile
of this distribution is 24. What is P(X 4.8)? .2421
4-5 The Chicago Cubs, the major league baseball team, just opened a ticket office downtown Chicago. It was
found that the sequence of arrivals of the ticket buyers at the office is a Poisson process of the rate of 45
per hour.
4 Within a randomly selected 4-minute period, what is the probability that more than 2 arrivals at the office
were observed? .4232
5 Suppose one ticket buyer just left the office. What is the probability that the next buyer comes into the
office no later than 1.5 minutes from now? .6753
1 If it is known that = 8000, = 900, P(X 8800)=p, and P(7400 X 9000)=q, What is 2p+q?
5-6 Warren Inc., one of the major manufacturers of TV sets in the country, is making production plans for the
coming season. Let the random variables X and Y denote forecast sales in million dollars in the coming
season for TV set models A and B, respectively. The joint probability distribution table is as follows.
Y
X 1 2 3 4
1 .12 .13 .06 .05
2 .07 .09 .08 .07
3 .05 .05 .11 .12
5 What is P(X2Y2)?
6 What is E[XY]?
7-8 The means and variances listed in the following table are calculated from the joint probability distribution
table of random variables X and Y.
Random variable Mean Variance
X 3 0.7
Y 2 0.6
7 Suppose E[XY]=6.60, what is the covariance {X, Y} of random variables X and Y.
8 Suppose E[XY]=5.40, what is the correlation coefficient {X, Y} of random variables X and Y.
9 There is this normal population with mean 40 and standard deviation 10. Find the maximum possible
10 Fifteen people signed up for the Taipei Super Marathon, 2018. Eight are Taiwanese, others are foreign
nationals. Everyone is well-prepared to win. What is the probability that at least 3 out of the best 5 runners
are Taiwanese?
11-12 The Chicago Cubs, the major league baseball team, just opened a ticket office downtown Chicago. It was
found that the sequence of arrivals of the ticket buyers at the office is a Poisson process of the rate of
λ=16 arrivals per hour.
11 Within a randomly selected 20-minute period, what is the probability that no more than 2 arrivals at the
office were observed?
12 Suppose one ticket buyer just left the office. What is the probability that the next buyer comes into the
office no later than 10 minutes from now?
13 Assume that we have a huge bunch of packaged tube socks. It is known that 44% are white, 17% are
black. The others are of assorted colors which will be called “colored” socks here for convenience. Ming
picks up, randomly, 6 pairs of socks from the bunch. What is the probability that there are no more than
2 pairs of colored socks out of the 6 picked?
14 There is this normal population with mean 333.4 and standard deviation 140. Find the maximum
possible value which is less than the top 11.70% of the population.
15 There is this normal population with mean μ and standard deviation σ . It is also given that P(X>
16 It is known that the 20th and 80th percentiles of a uniform distribution U[a, b] are 17 and 41, respectively.
What is the standard deviation of this distribution?
17-20 The shelf life of 8GB memory disk of a certain brand, in days, is considered a continuous random
variable X. The cumulative distribution function (CDF) F(x) is defined as follows.
21-24 The shelf life of 8GB memory disk of a certain brand, in days, is considered a continuous random
variable X. The cumulative distribution function (CDF) F(x) is defined as follows.
{
1 for x>30,
8x
F(x)= for 0≤x≤30,
30+7x
0 for x<0.
25-28 The shelf life of 8GB memory disk of a certain brand, in days, is considered a continuous random
variable X. The cumulative distribution function (CDF) F(x) is defined as follows.
{
1 for x>28,
x-12
F(x)= for 12≤x≤28,
28-12
0 for x<12.
Part III:
1 It is given that the random variable X follows a Bernoulli distribution with the parameter p= 0.53. That is,
7 Let X 1 , X2 ,…, X100be 100 random variables taken from Bernoulli(p) and they are independent. Let Y=
100
pdf?
(A) Must be continuous. (B) Must be non-negative. (C) Total area under the function must be 1.
9 Let X be a random variable, and symbols E and V are used for expected value and variance of the random
(A) E[3X]= 3E[X] (B) V(X+12)= V(X) (C) E[X+12]= E[X]+12 (D) V(2X+1)= 2V(X)+1
10 Let X and Y be random variables, and symbols E and V are used for expected values and variances of the
random variables. Suppose that Y=aX+b, which of the following is not correct?
(A) E[Y]= aE[X]+b (B) V(Y)=a 2V(X)
(C) If V(Y)=36, and V(X)=4, then a=3 (D) If V(X)=4, and a=3, then V(Y)=36
11 Random variables X 1 , X2 , and X3 are independent U[0,1] random variables. What is E[( X1 -2 X2 + X3 ) 2 ¿
?
(A) -0.46 (B) 0 (C) 0.5 (D) 1.46
Keys
Part I
1 .6103 2 .9793 3 .1112
4 .1539 5 .7642 6 .9778
7 .9699 8 .2877 9 .6719
10 1.48 11 0.13 12 -2.05
13 -0.23 14 1.83 15 0.28
Part II
.1867*2+.6151=.9885 8423 6 60
(38/49=.7755) (5.18) 0.6 -0.92582
(40-1.282x10=) .5734 .09924 .9305
.5651 500 912 Sqrt(133.33)=11.55
(.6321) (.18586) (.60653) 24.079
(16/17=.9412) (200/205-120/135=.0867) ((1-16/17)/(1-8/10)=.2941) 210/31=6.77
(6/16=.375) (5/16=.3125) (0.6) 23.2
Part III
1 (var(X)=p(1- 2 (-1792) 3 ((-1260) 4 (-110,880)
p)=.53x.47=.2491)
5 (5.774) 6 (D) 7 (B) 8 (A)
9 (D) 10 (C) 11 (C)
When n is larger, the calculation of binomial probabilities can be tedious. Work load is not exactly the
concern but the precision is. With the help of modern powerful computing machines, it is no longer a
serious difficulty. But in terms of illustrating approximating mechanism, it is still a plausible piece of work.
It will be illustrated with a plot to show you how to find the approximate probability of P(X=10) where X~
Bin(40, 0.32). The corresponding mean and standard deviation can be calculated and give us the normal
counterpart N( μ=np , σ 2=np (1− p) ¿=¿ N(12.8, 8.704). The area under the curve on [9.5, 10.5] will be
used to approximate the original probability P(X=10)我們用區間[9.5, 10.5]中常態曲線下方面積當作 X=10 之機率近似值,乍看之
下好像並沒有太大好處,讓我們看下一小題:
yn1
0.14
0.13
0.12
0.11
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0
x
Now we want to find P(12 X 15). It takes four solid steps to finish the job. With normal approximation,
one step is all you need. Furthermore, the normal approximation to the binomial is known to be very
reliable. What can we ask for more when it comes to approximation?
P ( 12≤ X ≤ 15 ) ≈ P
( 11.5−12.8
√8.704
≤Z≤
√ 8.704 )
15.5−12.8
=P (−0.44 ≤ Z ≤ 0.92 )
=0.4912.
ym1
0.14
0.13
0.12
0.11
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0
x
4-13 Binomial(40, 0.32). Find the following: (1). P(X 10), (2). P(12 X 15).(Exact. 0.09015, 0.48474,
Approx. 0.0863, 0.4912.)