Scribd
Upload
8 views6 pages

ML Assignment

Uploaded by

rravireddy728
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
8 views6 pages

ML Assignment

Uploaded by

rravireddy728
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Regress fon

Deftnitien Ragess ion is a Supesvised leaxving techniqyo ued


to orcdict a ontnuog stome vaiale (deyerdent Vatiale)
baged o ne 3) male poiedict) vaslableg ( rdependet
vaslabe)
Linen ge ion sprobably the most popub) em of eg egfon
aralyis te uge ef itp eaye - of -uge Gn phedicting ad foieatiny
Lineay Regoe ss ionj.
The linacen ession model poovideg a 6le
otia
dependent
Data points
soped staight line epOnting the
velatfonahip betueen the vcoicbe.Confde
tke inage:
Mathematialy we Gm ereRet alireay egies'on
y= ata,x +¬
Hore y =Deqendent boiabe Carget rdeperdent Ua fa bíes
x= Trdepernderst Vasiakle oiabe)
(paedct;
Vaoriabe)
a08- irsfecegt of the Itne (Sheg am addital cegee af feedom
al - ran Ygeion Geffrent (sale factßi to ach nput Vale).
The value toi x and & are teiing dtageto 1 lirn Regze scion
model epAeLHtatiern
Typer of Regres on
’Rolyromial Regesio. Gtend inan eg ion by tthg apby naoni
-al ation to the dota
’ Logistte Regrenion; Uced tà bincoy car?ration polens pedit
Rrg the pmbabilty of a bircey at Gme
keyCorcept
5 loss Fum ton ; eagres the en6) kotueen psredicked and akual
valueg. Gnmonly lged loss fnCtion, rcude Maan Bqiared toot
nsE) ad Mean Abjelte GycmAE)
Gradient Descent :
Gadient s uped to iniie tte msE by al Galating he gode
ht of tte Cot fumction.
update the oetficknto
A
eegion model uyes gradiet derent to
Of the lhe by aGdu Ciy the ast kumction Coefffient and
of alueo f
dene by a xando Sele ctton
ott
ther ttoatiely updte the valueg to ech -the nininum
function
’R-Squage nettodi goodne
R-suaed s a Gtatiria) method that detenineg the
-S of t
relatienehip betweery -fhe depor
Tt meagLIER tke terath ot tke
Qf O-lo
-dent ard hdegerdent vaabes n a 8ale
deterog the legg dffferece bw
Te high Value of Rseuohe
-the pedicted valueg and actual alue! ard bence repreeto a
Sood made).
Caeffiient of deteOingtn. GeffCert
IE is alo Gled a
of mattiple egoesior). Explained vclation
f8mla ?s
Total hstiation
APpIRaiongt
Reqesion is widely yed in vaias fiedt such af fenance
(paedicted tock pAkes ), health Ge (egtfmating patiert outo
-me), and maresig fle aeting Salet).
Noive Cla sihcation

is Baged on Gryels theler. and uged feh


fo stuirg clawfation
psobleno7
bahy is t aled Nahe Baye9
Bayey algiithm is oMplsed of tao widg Nahe and
The naie
Byee cchich @ be dercbe d ap:

the ouene
-ne fa eHtain Reota & Sndependent of
Df a erto featwres. Such ap f the Buit is ?dertifted o) thee
ed, sphex?al, amd
bayee of olor, shape and tagte then
Sueet fuit is ecogni zad an apple tence cach katge
a
Tdiidually onhibutey to identify tat tt s an apple wthot
cderendig On each otfet.
t depend on the pancade
Roueg beGrmo
-> Bayer : als Gled uped to detenire
of Gayels Thee8 eaye's Reule , which is
bypothety utth paið trawleche Ct depord
the pnhabitty of a
on the Grditioral po babi (ity plslA)CPA)
8imula is : P(Ale) Ple)
hee P(ale) is Pastesia) Potabiliy
PCo A) Ps Litelihood probability
wtirg of Nale ryes Clauifeai
The nae ott af the cauifiey asene all featere ttt e
frdependent of ach othey Ths SimpHg the onputation
Of the lltolihood PCI)) ag the procuet of he didlual
potabilitice op each Feate gven the clay.
The ster ane
’ Calauote Poit Bobabilitie,:
belorine the pui probability ftn each clax
-calalote itelihood ompte the 1Pkalihood of ach Peatoe

Gmpte Posteior Ponbobttiey Ue Caye's Theaey to akalat.


he
-tfe ppsteaið psata biiy fo ach cau
Claukyi Ash» the Clak with wth the Blyhet pollea) pahablty
to the rktane.
Adunttrainicgge data
af Nae layel (lauifieni
Ty to implement and perfons well euen oiik limi ted

ol nuna eate data s attumed to Gome n) normal

Nstsibtion
Sadun ntlee. of Nate Rayer claitei
Asume tat feahe ndepondort ,ohth may net alu
hold n a-wld
data
May usgn to e0 pkabilitey to useon eents, leding to por)
geretalization
-fepliGatiom of aive
Gmai) -Pteiry: Can?fey emaily ag soam
Sram Gmai) 8 non- Sam
based orn fatneg
rdlidualg ß
Crelit coing : Cialuate credit odHKRregR of
Loan apPn Val baye on
cweatte Gnditfon s
*Weatfe oyedlaion, Clafe Coeothe
Voh'oug -factoy
Support Vector achne (SVM):
SU a that
Canifieg data fnm to &i mÝe ffenent cayes- SVms aaea
forn other cariation algafttrs beauye tfey hesse the
decision boundaouy tat raxni2e the dlstamce hom al the
cayeg neaneet data poiot The decision boundy Chatd
by Suns is alled the axim masgtn hypeyplame, and
JM is alled the Marimun n clayfty.
ypexplane and suppot Vectors n sv)
m-dinenneral . n SUM , the hyerplar
sthe dectson brurday that epasatey dferet care.
The qoal s to maxiMiE the mogn, ohich s the dPstan ce
betuween the hypexplae ad the neoN t cata pointo forn eoch
clat (supn vectors)

-’ Support vectorsi Sup port Vectors e the data porots closert to the
hypeaplene. They ae Cala! n def fning the pation ard coiertatin
of the hyrplane
and the
Mangini The mas betweem the hypes pne
lr-the g2p betaeen
Support ecors A aagey mogin is Corldees bettey ag it
Gmpliey a mde vburt chrft

Snal masgin lage rongin


kernel cti Svn an efidenty petfem a non- lrey
claifioton ufngthe keonel tict ,ohich nplity napd
Snpup into hgh -dierg iara featoe Snce
Conen keney hcude poly roial ,and sada!
bagis functon RBF) Ferel
SVog aloith Df to type:
-’ lheas sVmi ohen the data pont aoe Rneasly Sepsple to
spaceg.dienioral -Algh n-ey
cten i todue8matig bioin eCagmition
nd Fnoge
tton, tet inwyed widely isisSVm Gehong) Appli
points
dota -theepasYote carifes
to SUN) (na mon- the uge 0e
oabe, lney not dota
s the When sUMi Non-Lhea ’
data Such clagity
the claits
to svn (Roea the e Ce
cdata. sepcoatle lneany alled 1sdata the clasjeg tuoo

You might also like