Professional Documents
Culture Documents
Christian Notes For Exam P
Christian Notes For Exam P
i
i
B A A " Pr# " Pr#
. Also from the
conditional pro(a(ility formula,
i
i i
B B A A " Pr# " - Pr# " Pr#
.
Bayes Theorem
" Pr# " - Pr# " Pr# " - Pr#
" Pr# " - Pr#
" - Pr#
C C
A A B A A B
A A B
B A
+
Epected !al"e
here are t&o &ays to calculate the expected 'alue. he (asic, most direct route is one of
the thin/s you ha'e to kno&!
0or a discrete distri(ution f#x",
x
x f x X E " # " #
0or a continuous distri(ution!
dx x f x X E " # " #
*ut theres also another &ay to calculate the expected 'alue that can (e faster, (ased on
the information you are /i'en in the pro(lem.
0or a discrete distri(ution!
x
x X X E " Pr# " #
0or a continuous distri(ution,
dx x X X E " Pr# " #
&here
" # $ " Pr# x F x X
, &here 0 is the cumulate distri(ution function.
%ou ha'e to kno& the (asic &ay to calculate the expected 'alue1 the other method is also
nice to kno& and can sa'e you 'alua(le time in the heat of the exam.
!ariance
he 'ariance of a random 'aria(le X, denoted Var(X), is /i'en (y the formula
, ,
" # " # " # X E X E X Var
Co#ariance
he co'ariance (et&een t&o 'aria(les is
" # " # " # " , # Y E X E XY E Y X Cov
.
Also, you need to kno& that
", , # , " # " # " # Y X Cov Y Var X Var Y X Var + +
and that
" # " #
,
X Var a b aX Var + and
". , # " , # Y X abCov bY aX Cov
he standard de'iation, denoted
X
, is simply the s2uare root of the 'ariance,
" # X Var
X
.
Coefficient of 'ariation! his is simply the ratio of the standard de'iation to the mean,
that is
" # X E
X
.
Do"ble Epectation
E(X) = E
Y
(E(X|Y=y))
3ar#4"53ar
%
#E#4-%5y""6E
%
#3ar#4-%5y""
7asnt on my exam, (ut you ne'er kno&.
Probability Distrib"tions
8ean, median, and mode
he three most tested pro(a(ility distri(utions are the uniform, exponential, and Poisson.
%ou also need to kno& the (inomial, /eometric, the ne/ati'e (inomial, and the
hyper/eometric.
The Uniform Distrib"tion
his is the simplest of the continuous distri(utions. %ou are /i'en an inter'al (a,b), for
&hich the likelihood of any point in the inter'al is 9ust as a likely as any other. he
pro(a(ility density function is
a b
x f
$
" #
.
he mean of the uniform is
,
b a +
and the 'ariance is
( )
$,
,
a b
.
The Poisson Distrib"tion
he Poisson distri(ution is used to model &aitin/ times.
he important stuff!
he mean 5
he 'ariance 5
he mode is e2ual to lam(da, rounded do&n to the nearest inte/er. 0or example, a
Poisson distri(ution &ith mean e2ual to :., has a mode e2ual to :. A Poisson &ith mean
e2ual to : also has mode e2ual to :.
Also /ood to kno& is that the sum of t&o Poisson distri(utions &ith means
$
and
,
is a
Poisson distri(ution &ith mean 5
$
6
,
.
)t /ets a little trickier if t&o Poisson distri(utions or more are in'ol'ed. A shortcut that
can sa'e you a si/nificant amount of time is reco/ni;in/ that the sum of t&o or more
Poisson distri(utions is also a Poisson distri(ution. 0or example, supposed that you are
asked the follo&in/ 2uestion!
A (usiness models the num(er of customers for the first &eek of each month as a Poisson
distri(ution &ith mean 5 :, and for the second &eek of each month as a Poisson
distri(ution &ith mean 5 ,. 7hat is the pro(a(ility of ha'in/ exactly t&o customers in
the first t&o &eeks of a month< he lon/ &ay to do this is to fi/ure out all the different
com(inations =
Case ) = one customer in &eek one, one customer in &eek t&o.
Case )) = t&o customers in &eek one, no customers in &eek t&o.
Case ))) = no customers in &eek one, t&o customers in &eek t&o.
he easy &ay to do this is to use the fact that the sum of t&o Poisson distri(utions is also
Poisson. So the sum of the Poisson distri(utions from &eeks one and t&o is Poisson &ith
mean 5 >. he pro(a(ility of exactly t&o customers is . +?@ . +
A ,
>
> ,
e
The Eponential Distrib"tion
his is another of the essential distri(utions. he exponential distri(ution is used to
measure the &aitin/ time until failure of machines, amon/ other applications.
x
e x f
" #
he mean e2uals
$
, and the 'ariance e2uals
,
$
a x
a
e dx e
a x
a
e a dx e x
" #
$
a x
a
e a dx e x
+ +
" " ##
$
, , ,
The $amma Distrib"tion
)ts /ood to ha'e a passin/ familiarity &ith the Bamma distri(ution. he sum of
exponential distri(utions is a /amma distri(ution. he exponential distri(ution is tested
'ery hea'ily on the exam, and there has (een at least one recent exam 2uestion &here it
&ould ha'e (een helpful to kno& that the sum of t&o exponentials is a /amma. hats
a(out all youll need to kno&, (ut you mi/ht /et tested on the /amma outri/ht, so listed
(elo& are some rele'ant formulas for the /amma. )f pressed for time, skip this and focus
on the (asics instead.
Bamma pdf!
x
e x x f
$
" #
$
" #
The Berno"lli Distrib"tion
Ciscrete distri(ution, the simplest pro(a(ility distri(ution, either an e'ent occurs, or
doesnt occur. A pro(a(ility is /i'en for the e'ent that the pro(a(ility occurs.
'
q p prob with
p prob with
X
$ . +
. $
E(X) = p
Var(X) = p(!p)
Binomial Distrib"tion
" # "
p p # " C p " # f
" $ # " , # " , , # &here
.
"A # A
A
" , #
" # "
#
# " C
Y 5 D failures (efore first success!
p p
Y Var
p
Y E
p p " f
"
Y
$ $
" #
$
$
" #
" $ # " #
,
%egati#e Binomial Distrib"tion
" , , # " , , # p " # f
#
"
p " # f
B $B
p
"
# E " #
,
_
p p
" # Var
$ $
" #
,
&ypergeometric Distrib"tion
Esed for samplin/ &ithout replacement. 0inite population &ith n o(9ects, k are special, n.
k are not. )f m o(9ects are chosen at random, the pro(a(ility that out of m, x are special is
"A # A
A
""A # # "A #
"A #
"A # A
A
" #
% # %
#
x % " # x %
" #
x " x
"
x f
)t looks a little complicated (ut once you'e &orked se'eral pro(lems, this is not too
hard.
%ormal Distrib"tion
Continuity correction factor for (inomial or Poisson or uniform approximations!
,
_
,
_
+
+ < <
> . + > . +
" > . + > . + Pr# " Pr#
" "
" Y " " X
Bi#ariate %ormal Distrib"tion
"" # # " # " - # X E X Y E x X Y E
X
Y
+
, , ,
-
" $ #
Y x X Y
Lognormal Distrib"tion
,
_
y
y
y
y Y y Y y F
ln
ln
ln
" ln Pr#ln " Pr# " #
,
ln
,
$
ln
" #
y y
e Y E
+
,
_
$ " # " #
,
ln , y
e Y E Y Var
Other distri(utions on the sylla(us include the (eta, the Pareto, the Chi.S2uare, and the
7ei(ull. ) ha'e not presented them here.
Marginal Density
X Y
dy y x f x f
-
" , # " #
'rder (tatistics
" # "
" X
x F x F # " C x #f x f
"" # $ # " # " $ , $ # " # " #
$
" #
Conditional Density
" #
" , #
" - #
-
y f
y x f
y x f
Y X
X Y
X Y
dy x y f y x X Y E
-
-
" - # " - #
Moment $enerating )"nctions
&
X
(t)=E(e
tX
)
&
aX
(t)=&
X
(at)
&
b
(t)=e
bt
X,Y independent 5F &
X'Y
(t)=&
X
(t)&
Y
(t)
&
X
(()=
8B0 for *ernoulli! pe
t
'q
8B0 for *inomial! (pe
t
'q)
#
8B0 for Poisson!
" $ #
" #
t
e
X
e t &
8B0 for Standard Normal!
,
,
$
" #
t
e t &
X
8B0 for Normal!
, ,
,
$
" #
t t
e t &
X
+
,
_
$
$
Note that (ased on a comparison of the 8B0s for the exponential and /amma
distri(utions, its easy to see that the /amma is the sum of n exponential distri(utions.
Goint 8B0s!
,
_
tY )X
e E t ) &
Y X
" , #
,
( ) ( ) + , +
,Y X % #
% #
% #
&
t )
Y X E
+
A couple of other important formulas that can come in handy!
" # " + , #
,
) & ) &
X Y X
" # " , + #
,
t & t &
Y Y X
" # " , #
,
t & t ) &
Y X Y X +
Chebyshe#s Theorem
( )
*
* X
,
- - Pr
Benefit Distrib"tions
E(X) = qE(B)
Var(X) = qE(B) + (qE(B))
,
Miscellaneo"s )orm"las
K
th
central moment of X ! E((X!E(X))
"
)
Correlation coefficient!
Y X
Y X Cov
Y X
" , #
" , #