Beruflich Dokumente
Kultur Dokumente
1
Course Topics
Overview of Industrial and Systems Engineering.
Probability in a finite sample space.
Conditional probability and Bayes theorem.
Discrete probability distributions.
Continuous probability distributions.
Probability for systems with 2 or more random events.
Statistical sampling and confidence intervals.
Hypothesis testing.
2
Statistical Independence
PB j PA | B j
PB j | A = k
PBi PA | Bi
i=1
4
Properties of Probability Function
5
Definition
7
Definition
6!
C(6,2) =
2!(6 - 2)!
= 15
10
Negative Binomial Distribution
pq x-1 x 1
P(x) =
0 elsewhere
12
Hypergeometric Distribution
Consider a discrete sample space S where
N = # of elements in S & k = # of events A in S
Randomly sample n elements without replacement
from S and define the RV
x = {# times an event A is selected}, then
k N-k
x n-x
C(k,x)C(N-k,n-x) 0 x n
P(x) = h(x; N, n, k) =
= C(N,n)
N
0 elsewhere
n
Similarities Differences
Sample n items, x of Binomial:
which are event A and Bernoulli property
remaining n-x are event satisfied (i.e., events
Ac. statistically independent).
14
Binomial vs. Hypergeometric
Binomial Hypergeometric
Distribution Distribution
15
General
Hypergeometric Distribution
Consider a sample space containing N discrete
elements. Each of these elements is classified as one
of J events A1, A2, ..., AJ where there are ki of each
event Ai and k1+k2+...+kJ=N.
Select n elements at random without replacement
and let the J RVs be defined as
Xi = {# of times event Ai is selected} i=1,...,J
then
C(k1, x1 )C(k 2 , x 2 )...C(k J , x J )
P(x1, x 2 ,..., x J ) =
C(N, n)
Note that order does not matter & x1+x2+...+xJ=n
16
Uniform Distribution
1/k i = 1,2,...,k
P(x i ) =
0 elsewhere
17
Poisson Distribution
0.2
0.15
0.1
0.05
a b x
0
x f(x) measures the density
a b
of probability between 2
What if b = a? values 20
Final Note
21
Uniform Distribution
1/(B - A) A xB
f(x) =
0 x < A, x > B
22
Uniform Distribution
If A a x b B, then
b 1
P(a x b) = dx f(x)
a B- A 1/(B-A)
b-a
= a b x
B- A
Note that
P(axb) = P(a<xb) = P(ax<b) = P(a<x<b)
23
Uniform Distribution
0 x<A F(x)
x - A 1
F(x) = AxB
B - A
1 x
x>B A B
Similarities Differences
The random process is a Poisson:
Poisson process. Interested in a RV X that is
discrete.
represents average #
of events per unit time Exponential:
or space. Interested in a RV T that is
continuous.
A normal RV is denoted by
XN(,)
where
= mean & = standard deviation
and the pdf is given by
-(x-)2
1 2 2
f(x) = e - < x <
2
Also know as the Gaussian distribution.
27
Normal Distribution
x -
F(x) = FZ
28
Example
z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
1.9 .9713 .9719 .9726 .9732 .9738 .9744 .9750 .9756 .9761 .9767
2.0 .9772 .9778 .9783 .9788 .9793 .9798 .9803 .9808 .9812 .9817
2.1 .9821 .9826 .9830 .9834 .9838 .9842 .9846 .9850 .9854 .9857
29
Special values for XN(,)
+ - - -
1. P(- X +) = FZ - FZ
= FZ(1) - FZ(-1)
= 0.8413 - 0.1587
= 0.6826
Approximately 2/3s of the outcomes from a
normally distributed random process will be
within one standard deviation of the mean.
30
Special Values for XN(,)
f(x) 0.5
0.45
0.4
0.35
P(- X +) = 0.683
0.3
0.2
0.15
0.1
P(-3 X +3) = 0.997
0.05
x
-4 -2 0 2 4 6
2
3 31
Weibull Distribution
A continuous RV X has a Weibull distribution
if the pdf of X is
-1 -ax
ax e x0
f(x) =
0 x 0
Probability density function Cumulative distribution function
where the constant parameters a > 0 & > 0.
-ax
1- e x0
F(x) =
0 x 0
33
Bivariate Distributions
34
Schematic showing joint, marginal,
and conditional densities
35
Definition
Y 0 1 P(x)
X
The RV Y & its
0 P(0,0) P(0,1) P(0) numerical
values.
1 P(1,0) P(1,1) P(1)
38
Example
Y 0 1 P(x)
X
The RV X & its
0 P(0,0) P(0,1) P(0) numerical
values.
1 P(1,0) P(1,1) P(1)
39
Example
Y 0 1 P(x)
X
The values
0 P(0,0) P(0,1) P(0) for the joint
probability
1 P(1,0) P(1,1) P(1)
function.
P(y) P(0) P(1) 1.0
40
Example
Y 0 1 P(x)
X
The values for
0 P(0,0) P(0,1) P(0) the marginal
probability
1 P(1,0) P(1,1) P(1)
function of Y.
P(y) P(0) P(1) 1.0
sum P(y) = P(x, y)
x=-
41
Example
Y 0 1 P(x)
X
A reminder
0 P(0,0) P(0,1) P(0) these are
probabilities
1 P(1,0) P(1,1) P(1)
& must sum
P(y) P(0) P(1) 1.0 to 1.
43
Definition
f(y) = f(x, y)dx
- 44
Properties
45
Illustration of Covariance
Y Y Y
X X X
47
A Problem with Covariance
COV(X, Y)
r=
VAR(X)VAR(Y)
Note that: -1 r 1
Also, r has same sign as COV(X,Y).
48
Whats It All About?
Consider a random process with RV X.
Probability Statistics
The type of distribution, The type of distribution,
and are all known. and are all unknown.
Use this information to Sample the output X then
compute the probability estimate and from the
of various outcomes X. sampled data.
Use this information to try
and determine the type of
distribution.
Called parametric estimation.
49
Central Limit Theorem
Consider N RVs X1, X2, ..., XN where
E(Xi) = i & VAR(Xi) = i2 i = 1,...,N
52
Statistical Sampling
distribution values
parameters from a sample
for the actual of the actual
population population
mean x
standard
s
deviation
53
Once again, and are
Notes parameters for the
actual random process.
Some
. notes about the RV x.
Since it is a sum of RVs, it is an RV too!
Sometimes called an estimator or point estimator of .
The CLT implies that it has approximately a
normal distribution with mean of
E(x) =
and standard deviation of
Estimate of
gets better as
x =
N N gets larger.
We dont know the distribution of X ( and unknown.
54
How Good an Estimate of ?
fX x
Sample sizes
N<N<N
LargerWhy this
sample size N
behavior?
implies x more likely
to be closer to .
x (lbs.) 55
How Good an Estimate of ?
x s/ N
If N is large, this is a good estimate.
However,
if N is small, then will get a more accurate
result using the t-distribution rather than the normal
distribution.
s 565.6 s 483.2
X = =178.9 lbs X = = 48.3 lbs
N 10 N 100
Thus, a 95% chance 3077 lbs. Thus, a 95% chance 3156 lbs.
is within 357.8 lbs. of the is within 96.6 lbs. of the
average U.S. cars weight, . average U.S. cars weight, .
P(2719.2 3434.8) = 0.95 P(3059.4 3252.6) = 0.95
58
Whats a Confidence Interval?
59
Definition
61
Recall This?
62
t Distribution
Find
a. t0.025 when v = 14
b. -t0.10 when v=10
c. t0.995 when v=7
64
Confidence Levels - t Distributions
65
Hypothesis Testing
66
Format of the Hypothesis
67
Format of the Hypothesis
68
Testing the Hypotheses
70
Hypothesis Testing Errors
72
Minimum Significance Level