Uhit:2

Page No. :
Regresson Date:

Regnessian

Regnessian ek statisticaltechnigue haiska har

USe Contiauaus atahles ke beech tlalionshi2
ko odel ke kiyesta hai sme
ek depepdent \aciable. ekya ko

independeat vaniables(iaputs)
ke
Zyada
prediat kiyaAtahai
lels
kamo haiki
tumhe padit
kisi student ke kitne
aayena (depesdent aiakle
)
ka Use kar
sakke ho

Linaar Regaessie Sinaple

Somple
ak
for
tioal
stactis nethe hai do vaiablee e
heech lineax aelahonsbip ko model karta hai
ek ek
independent Cnpat) degenslent
(auau

Epaban

Depeodent Vnisble autpu)

idapandeni yaniahle(input)
e/ slape (aelahanshipstasngh)
Intersept wbere
y
line crSes
yxis
)
Váy Line of
Regressio
Deperctnt
Indeperdent Var
 Page No. :
Page No. : Date:
Date

Inear Mulkiale lineanr Yaiable(Output)

Milkigle
ki
tah
ko
hota
karae ke
ek depsndeni V9Tiable x2, lndependent
Vanables pauers
predict nuligle indepencknt
VanableS ipvalved ho linea
nelahaoship foctors ke saath
model karta hai

Fquahion

Dependent Vansiable(aatpu)
Indepencdest Vamchle (lnpu) Clasification Algsrihn
Mulliale
Coefficient Ralatiaeship
shengh hetwen hta
technique ha
sanables) Clay,ifcalee
alyoahm
categories(clases)me diyde karte
lineof regTession ba ka discreee auteut
ko prediit karne ke
Dotapsints
hai aise whe the
spana
et spam

Lagistie ek linesr

Paly sanial Regesin Outene paechict kar to hai Lekin

Palyosmial ek exfension hai egtesstn ke ke bhi ek
ka ysme independeat classikicoin
Lineax kaake Yeb
ki alasificgthsn ke ise hota
hinsy
Vaoble, ba data uteomes do hai
CUNe ttingkiyayala
ho
hate hai
lines ya Yas Quy Ne)

sMe
sigmad uochioo ka
adic value ko o
ke haech map katg
 Page No. :
Page No. :
Date
Date:

Forinla
Assune aiihute
Sigmad fehian
TRme ( clickedso Ad CY
P: dowr N= Up
No
4o23 Yes dassA
IG = N
PN PtN) PtN
39.45
Yes
28-15 No
No I6 =

DecisiCnee
Decisioo ree- ike stauctue hata
hai isme decisinns Seuence
questins Condiiss se basis
Deus
ata har Hax ncce ek featige ka
epreseat karta hai Doun
hoanches nesults
Acpaesent karte hai last mein lec
final class lables ko epresrnt karte
hai
E(A) = P; +Ni
Ase Competiton ype
PtN
XI(PiN)
Yes.
2 No
No
ilola)
Douwn

Yes Daun 340

Yes

Ne
Tes
 Page No. :
Date
Page No. :
Date:

1X

O.34

0+2

I(Ne)
=

EA)
Gaia
(Age)

Doso
Yes
P+N
Ne

3
TM 3t2
3+2

I(S/)=
ECyes)= 242
242
 Page No.

:
Page No.:
Date
Date :

3 vahue
142 K
yeh daside
neishbau Consider Noscest ka
data pointssingut
space sobse hoi Msually distance
034 like Euclidean distsnce ka b)e hota

Euclicean Distane :
where Chseved Vahe
Gaia (ype E(T) Centid Vae

Haght
2
chsters k,l k

60
Age s5,22)

Down 72 We hae
Up \85 0

ype

Daun

Tigure i DeisienTree

i)k-Nearest Neighbsrs (keNN)

kisi
in csse ke
ek Iazy lenaingslilhn
k-NN hai
bhi karne ke
dataçoiat
sske nearest axistigitcheck ko

kaota hai
 Page No.
Page No. :

:
Date:
Date:

Supot Veco Machioe(SVM):

Distance lena
S VM ek
Supervised machine
algsothaa hai
diffexen classes
hypplageka
KO
kante
Yeb
Seprae Kame ke Po
hpplane hne hoti hai
ke
classes
Gueth
data
sepahon
beech

non- inear
eate
naximum
karti Lke to
Linear SpTable

"kerneltaick seprahle
Ka Use kagcta bai
Ohgher dimensions
nsse dsta ko (wNaive
pacin transtorm kaske inecn
septabie banaya Sakta hai
Nae Baes ek probabilisticalasifierha
theogem pe base hota hai Islka

ModelTaining
yeh hota hai kihM
feature
icdepeadent hota hai hence
OCincle
Predisthon Neus
Yeh prnbabilhhes ke basis per prediaians
ceta ha efficicnt fast hata
Otput hai datase ke
especially
D:tecea versins mive
Rolo Ynargin
this Qxople There hote hai aise
bae ccsifies
two Gaussian Naive
MultiniNaiNe Bayas (Condisusu)
COuad letal Dne Csaitganiaal dal)
pey Gaussan Naive
Hypesplane also kobun Paes Gausian aine
ane
clecisisa brandey date ke Coahsuous (eal-
aied
bhi ko data çaiol
hai s
distbubicn
hata
hax
hi ki
feature ka
oske hon
classifica ke liye ture paedislon data ka nonmal(Gaussan)hota hai
ke e hume cecisitebaonday ka naximum distaibuhoa ek
fuibu kagta
ha
Matlahbell
sbape
 Page No. : Page No. :
Date: Date:

b) Multicomial Naive Multinscaa

Vi) Nevnal Netusok
iscete data ke
Use hota hai Iiye
QSunt-aced data ke Neural ek
Netwesck bialcgikaly ispined algomthoa
Yeh tab kaam hai b featue neurons kote
epxesect kade hai ki ko Value kitnibLr he iotornaton ka process Kame
(ie frequenty Couct) Har ioput Ke sath
korta hai fic actiatian
Epaahe prehict karaa hai ki ek email functise apply kaxke Culput
base

specifie keyuordi (ree cffex)email Neuasl netunnk

mltile hidden ho
coamplee nela henebips
lc) kane ke ye kaanm karti hai Yeh
Bescoaull; Naiye
Besnollk NGiue Payes bingry data kaafi hoti ai epeSall
(o/4) ke natural
Cspecially e cogihihon
kia hai Isome featires ko binai me
Variables ke treat kiyo poecesig
ata
hai aise
ka RSsuomptia heta bai k Perstor
nance Measunes
featusres Sigf tuo outcomes hete
posibe
bai
Deasuxes kaaf mpaetant de
hun clasificahiaaalaath
Coespaisan Table 2 kate ka
hai
Bayes Versjor DataType Use Cqse Example
hoi
ka chalta
Naive
pste
Gaussian ContinuoUS Nypárcical Features Age, Incotoe etc kaan
aha ha
Letis disciss
Mulhimcnia) Discyete fequercies
Berroulli Bintay Boolean featuresUdPresense

P(Success)= P. PCe)=pto-p) Ascuacy

Pxeision
Recall

failure
 Page No. : Herronenie Mean 2y Page No.
Date Date:

iCoofusionMatrix i TP
TP † Fp

NxN matix hoi hoi

ncdel ke predician esults. ko breslkchu Aecal neasiges kanto hai ki achual

eleneals hate ha pASItivÉs Casts me Se kiine cooTcly

Positive Negehve Actal hue
TP FP
frtiet
Neghe FN TN TP
Recall=
TP + FN

Fe Posihve Mael Cutecly pasitireclass ko

pomaict kia
(Ahn Rosihive Predictel+
cal
pecisiaa SurTehai
ka harnmanic hoto

Neaa tive Mdel ne Yeh ek Siore hai b paecisinn

Curecty negnhive clas ko balanced
medict kial Aetial Negaie PredictedHegstie rerall ke beech ho

False Psiive : Mode) ne

gakt poaihe peclictkiya(khual
Agatis
Paerisianx RecalL
Predite Posihve)se TYPE 1 errrKahtehat Pecision4Reca

False Negatie i Medel ne gplat ngalisce pradlct kyal Ashal boie,

Predicted Negatie)se TYpE2 eroY kahte hole TPtFP

TP
TP4 FN

PtFP 1P+FN
Claicatioageeunrscy se tuM yeh
koocte ha ki kitni paredichians T

total se
paedichons (Fe)(TeEN)
TPt TN
TC2T t FP +FN)
Accuracy TP+ TN +FP tFN
2TP
Precsiba measuxes paelicted 2TPt FP
u)Precisioo hai + FN
me kitne schal pesihives