(A487 Ch 4 Bivariate Distributions of the Discrete Type
Sometimes ve vant to conifer moe than one randomvariabl, ve vantto loka the
rebonship between wo variables.
‘We want find the best curve to describe the rebtionhip.
‘campe-what the ebtionship between the highschool andthe ACT. We can we the
rebtionships to predict thd varable-porsib GPa mcolege?
So bt’ foray define t-te and Y be two random arable defined ona dict sample pace
Let denote the conespondingtwo-dimensomalspace of andy, the random arabks ofthe
‘cate type The probably that)=xand Y= denoted by fh = PK, Y=). Ye function
4s) called the ot probability mace fuction (ont pn of and Va ha the following
properties
A) Ost yisaa
8) Sasyes BF C9)
9 FLOM) € AI= EoeyyenS fey) where Ais a subset of the space S
Example
‘Suppose ve toss api off, foursiled die, in whichone of the dike & RED and the other
ie auack We't et
+ Xz the ouome on the R&D de = 2.2.38)
the outcome onthe BLACK
‘what the potabilty hax tales ona yoru vahe x and Vtales ona partcuarvale y?
“hat what any?
Solution
{sta we fave to inthe caze with one decree randomvarable inower tind the joint
probability disuibuton” of XandY, we fst need to define the support of and. Wel the
suppor of xi
5:(1,234)
‘ad, the suppor of Vis
S=f1,234)
Nov ie tf denote one of the possi outcomes of one toss ofthe pairatdice, then
‘certain (12) pst outcome, asf (2,2, 3) and 4). ve continue tenuate allot
‘the posible outcomes, we son ste that the intsupprts has 26 posible autores:
5=(Ga (8, 9,4 (23), (22). (28), (32,033,388), 8,)(43,(44))
Now, because the dee are fa, we shoul expecteachat the 26 possible outcomes tobe equal)
el. Therefore, using he css¥alappreach tas gnig probably, the probably that equa
any partcub Value and Vequak any pac ur value, 64/36. Tat, foralley} inthe
supports:
Po een
secause ve have Memtified the probity foreach), we fave found what we callthe joint
probability mass function. Perhaps, ts ot too surprsing that the at probably mass urtion,
"uhh tpl denoted as fy, can be defined a formula (swe have above} 25a graph oF
352 tbh. Here's what our iat pm woul hen abu forme
‘he propertzs ofa pf of one random arable (a the mais) ald @ marge pat
4
et formal defin-et Xan ¥ have te pint as futon) withspaceS. The pt of 3 h & 2
sone wnchicalthe marginal probably mat functon of, & define by 90+
UhisE,f@y) =e are,
‘where the summation talenoveralpssbleyvaes foreach genx inthe xs 5,.th eE/G9) =PO eres, 20
Where the summation étalenoveralpossibley ales freachgiven inthe spe 5 & eu 3
“The the summation & overall) in witha gen vale Siar, the marginal probability 30 2 30
mass function of, defined by
Hi) = Ef 9) = PWT yy eS,
“The random variables Xand Yar independent tf P=», = ¥)= PO=A)PIV=Y)
Otherwke Xand Vare salto be dependent 6 tots
‘camp Lt the joint prt of and Vbe defined by thy! = 52.
Thenty)= + 8 12.3
% ® %
2342
SY, y= 12 Mote and ¥ are dependent
be hy = 70
Lette pint petit aad Ve defined
= L2 y= L284 Find the marginal pmf of ¥ and ¥,
‘Then fed 90<¥) rixeve3)
Sometimes the random vat
es are nated by 2nd:
es eeu) OF ea ty
Now ets fnd the variance
‘cample: Ther ae eghtsimibrchps na bovt-thee marled 00), to marie 0} and two
‘mated (0), apd one marie (3) 4 phyersekctschipat andomand & gente Sumo the
‘wo cooniats indoles. x and ¥ represent those coorinates, thera prt
2 SS ley)PEY
“Thus E(X-+¥) = O(3/8) +2(2/8) +2(2/8) +214/8) 0
hye Qe vite x= oad
Now lets extend o thee varbls-ft the hypergeometric
-anotherexample Corser poputionof 200stulemts who have jst fashed 2 fist couse in L\
‘akuls ofthese 200,40 have earned As, 08, and 3006, 0,orF. Asampk ofeke 25 ahenat | So
‘andomand without replacement from ths population and each sample as the probability CS) J
"ati sokcted. et be the nemberef Estee, Vike vember of stent 3d 5-1 e S35
‘Then ve low thatthe mara pf off)
‘02 # GK) 50X and’ are dependent
Now lets do this witha binomial dis-we have three mutually exclusive and exhaustive
‘way tor an experiment to terminate: perfec, "seconds, and defective. The
‘experiment is repeated n independent time and the probabilities
"by": =2-pc-py X= number af perfect items, ¥ = numberof seconds, andZ=n-X-V= number of defectives,
‘The probabilty of havingx perfects, y seconds, and n-x-y defectives is
PAPY, (L- e- Py)" acne at Pye Veyjeanbeschived t
dif ferent ways
Cleon -y) = ST
So the trinomial pnt srr PDs (L= Pa By)
Wher sond ys oneptie eget oy 4
{sample nanfacting certain em foun hat inno profutonabout sso the
item a god oms, 6a sents andar detective, tachouranonli inspector
bere 20 eset a and fn the probity nth samp se of = 2,
atleast seconds rat st defect tems an dove
SoktAs (yx 2ory22} then#la)=4- PA} =1- Pics Oor1and Y= 00r3)
_ 20! ¢ Js!
ay _ (os) Coif 3° rpg ag
ra (Olan (asf= 30 (od alla).
gl
Now t's ty another one-A manufactured tems csited 3¢ good, second, of
Aetectve wth probabiltes 6303/10, and 2/10 respective Fiteen such tems
Bre selected at andom fromthe production line et x denote the numberof
food Aen, Ythe numberof seconds and 25--Ythe numberof detective tens.
Find the jolt pratt ard, iy),
Find wx=20,¥= 4)
Find wx 32)