MODULE -2

Model Deaign Cloius :

ob as
We can wite te otnt

-) baco

that modol ats the data lettu Hhan a

hagpen that

dubbuen:

Hhe data

Mooel 2 it to ees pom con itth o:4 Gis

and othd
 tilclihd
Etimaton: ieahood Moim *
M
and caltalate y Can
Ni
Um
X) E NN lm
N
to that &aysapoah This
mamnts Hestmate
E(X)
)
onpooch Simplut
at todit Cawe
th tar the Vale -’
data the
Estmation: etdPoom *
he
ale the
8lo Consicy M
1b nt svatioa
A

LA) =

,
 0 Z0
lilatihod, log
and To
iz) 2
2
ND
(2r1)
e TT
invtdee wab, cOvaione
POf
E ( e
hauslan: Nltwalate
D Log
l) )oga (leosy- Log Ciciheod
)
Pxl,
lihoed: le to
/8/2023
 )
n=|

- Cov(Xi,,)
= E(X;X; ) -E(r)EK)
Betal a, b)

b)
TT (1-;)

Ra
29/s/2023
C(x')

Model 2! 4, , (B)TAl,9)

e 1000 sanp bsavatoy 10 batch

oNda a thenig Qat
B, , Batoh man
2
Bs, S

Bo
Statste
N

leaizatb
e orution: at e mean Rutmatead n
he actual man
 PSTIMA TORS
Qauat 2: u he eataatd pohabm ates
Oxpechd Vale

hen

Bias cattnate

qos ’Torqet o utmate.

bìas estimate <0 Aluwaye undo etuahu

unbia 2.
foa Bunouli

e(xi)

ofos Bhouli d)t. MG M

aays tan

N
E(K,- j)

-EXX; +XXi +X X;)

N

(ot

Blased.

N
E )

Cws teut umati

Unbied timatd

j
* Bayealon Pakom otoh Etimaton :

Modul

Py=1) 0.49 POY o)Dol

P(XIY=1)-0 0S P(X|40) =008

X 24 and hen |T:

Conid aMothe xawple

X= 90 Hand 10T
Px|y1) =00004

we conhruosly updote the model!paimabo lo Ba

 No. Date

21)g/:023

9 e dla Dlla dit at p.

to 0, Qud
9 Bet (a, b)
yptpahabs
Maiemwm A-Postaiasi tmate

Posteish

P{9/x1) Pix3/e) P/S

a iopßtekl kipu,
with 9.
a cont and doed not
b

P(ix/e) =
TT e" (-)

91-0) lo,b)
Cout BBnouli dÀt.
 ta (1-potb-1

Pla,b)
N, ta
(I-9)
p/,tae, N, tb ae)

Beta (N ta

e Beta
Plioi fo Barolli.
Beta ls a Conjugate

aud aamma ale ao conj ugats non

Prissom aaus ian dut Vaiane
stmate
Modal
Data

22

N/ X

whe u cn e hen w
we ae us celan about
choge a
/a023
RECAP:

posblo

We wont Postooh aund i n same fomily.

P(ex) P[x P

gomna a, b)

aposto:a (u,D) d(u,T, x). piel)p,()

Same
whih mdiotes that he pogtols oe n
 dist Post
LPKc|x:)
|x,) 2P(Y;= )
bte
clayed nmalC’obal
(CONT/DrSCPEE)
(D)seETE)
Pahameth as dloni Gays Naire *
Estmatien:
Sla023
to C-D
Vn
)
D(Zaye)
~U,)
then Exaple:
X;
too(l-a) (Uxs,(X)
the as
ound codn of
ie. totahqt9 the
wstHAtn L/Y) oud ly) stotate tus (usid
t wehidh tntval Confidante *
 Poxtoio
Clas conditioual

Y,, x\ aRe all Savdom val

In tiabig pha, we ty to
Lile tbathg,

tst

Nare Bays lasu

D

ie X, , J.... X, asmd (idapatot

Dou,
2,
, ’ ,2
’,2,-)
he ndu axtmiee
,Maxtle to Iagoal ouNow
D N
MLE.
appiach: Beut
=c) |4; T(X;=
Ik)
M;
Yi)e P;, (x,) - Polt Jot
N
and,
M
doltniton Paam
oto
of mwm cal
ouoixt cateqdi have can We
u L a coutatut optwi 2aton .dolons
C
GM

ou0 we &tmpy aleae pohal doivates ad

-Zs(-) dlog , 0

it

Stnily, hawing ith atlbute a k

for Bayalan anoly, we may put dihbt p
se dstouttoy

Pl0,}I{x,vy) aPix,43)e,7) PLLO, m3)

Param Pe

mulbipicaton with
P{x,Y} 10,73)
Ond olve t qet
Ngy t -Pgs )
,Me t )
w/023
A

whue w

N
X;Y; )
OLS Estmate’

e
{aachtal seonalis.

e s a Randon voh, ew N , )
w'x s fied dotministe
z) N(+ wx, r
MLE

T
 22

Was = WwE

aend the peditted ne, th Rome waione m.

We ue it when we hae lome pu buouege about w.

usthn a phele o Radis3P

D

in

UN CORRELATE.

Nlo, P)

P(wliy,v})

Plw) wNW,V, )
 IM

-
2

D
Hule , =0 w,= 0
-

the luution

-(
Wos

Pot oel

the waine
 heMaes
aelo hem, mako Meuhon,
hen al ned done we
Suppou
12)4/013
txV,x)
thatoytatn theomt ts e lan Dval,w
thon
heoem!
postaie. modt
o twhole the but
mE
Unie mode, 9nbaud Prodichim
not
DPast
data
P(w)D)
duo P(Y|D,w) P(91e)
PRediction
We dyfine w vectoà as
drnat whe multlrotis

Pios ol andn:
No,py) and

Postag Pw)D)a PD|w)PyPin)

5
P[w) vyil)
Snu vand n ae imdipndat,
sfab and pike p

(1-n3)
D

non -Rs olonty

wh

odnate A t : Dn
w,o)

Tu convigznte s, )