Infermatim Aflr roesseing the deta, we

tnformation .

There in a fneline diff. blo thoe.

- dato Mining Maehine Larniry
/es7

R0main
bUNBatasoienee
dovio
dorluhment
researcel
Logieul statoent bauh on etain asumtionk

-Thore ort a tyfeu of doto t quali tottve /eadagarisal

1quantitative
4aencle
tContinuous
torudinal
 Aoguitid prakatsiy

Modele
valilktien
ata viualirati
BAR
soatternd Plot istoqrar
forest Cerwe
Pix plot
KM CuNe

Aota claning
O
Normalixation
remeval
value hend lur

4
Thare are, tyfer ML modela
)Gubaovixad Larning (( tabelig pruent )
abrent )
in) Reinfotmt ) baadon fecdbaek,medil liarn)
fon xome éate
i) Seminuberglkd
2n xujorvind la
buf in tnauerovid
mcaning
1)
labeling in frnl,
 netoork) Advenhia Gunuraktve (LonY qAANL(STM
mmonY) tenm Shont
Neoork) nlua Reeunrunt RNN(
Netoonk feruoerd fd
Netwonk) nltural Arotfieial ANNN(
tros Percep +
forert dom Ran
Treesisin
Bap Naive
bouen) neih Neanert Kald KNI( N
Maeline) Veetor hor (Su GVM
Modole valuex)
bravidox okjut
s the of
aþartieular etimates
a Rgretion
(4 9)
ealngarieuitjeet latiafioton eD
, omt into -ML
metbnd diviaion xberifie
of tukes
of are Thtre
A 7
 slatistes | Oth an 094

Average
Avgof midle vae atlen xonkong .
Ifrequent,halue
Moxt,àauircing
Hlandard deyiatton

ANarciane; TVarian eo =g2(a- u) 2

- SD. nabreUntx Variablity of dala boinds aeruss tee

nLan how MCt Varionee refrRonds variability of
loe data aeros mean .

Grmal data uted for ’ T6tal datais ealle

oallid tofulation.
6

Cc n Exy- Exsy

4in a paramthr wd to miaxune te relatibilay

 torclaled it ee is 1, manla ane -voly cornelohed
’ 1{ ee ix o, uere ix no rolation

8Ronge- MaK- Min

Rreenile

Linar Refrinion : (ys mÛge)

Conftdenat nteroval

0 Gausson tueve/rimal urwe/ ell turev i

+ Sizeikalin : tikusion of data

oharoslucwies of Nlooma steibukion
forfutla dizhribuded doda man: mutians mode
O tor
Arua unden tu cutwe 1 unden norUnaly
diztrihuted dak
 2
2
20

lt6s68/
S

355 94.7.

clnieal
Adminixtratve ( Nedieal insurano)
Giomieirie
Geno nie (( Gonss
5umegra7iel oundor, loeakiod
Rexuarol ( data elalee to neeaneh )

givon hy ote)

data (eT Ran )

EEtehronie eaeane Reeoad
-( Blood pevsure TNOe- To overeONe
ofeatuoR overttting, no. of
of roue/2)
 Nlon- taramthrie mettad (xtoat)
Analysixot Varionee- ANOUA
Parametrie motiada, to llaos neromally distibuad dot.

M
730. J &claia

’
reeomnended

bung
Prábability value of mll hyhothusis meei true
IH ix alo ruferud a erltieal vale.standand valueof:

doie dato elaning

O 0-05
@ata Umdiraton
dota tandariation
Voriable
0- 02

Catgerica galut enosding

miking value tandler
 Mork
80
&
87
19 92
85

87

make -the data in eerfan

’ It in toAommundoo to
range. (K)
17 7 seorting mttod
) màn- man mormalizatim ( ' , X-Mmin
max min
aota otandarolization
normal distibution of data in
’4 ix wd fore
ick the mian qotl be og stanland deviaktn

Min- mek norumaliration cd inttit

One hot
D, Eneod ing (ud ton numinal data)
1) Jobelle onco d§ng ( wd tor. ond inal data)
Misng volue tandlere'
i) Embdded Mittad
 i) peA Cfrineibal tombonunf Avalyein )
) LDA (Lintar Diseriinent )

Dforard feature ealeohon (EFS)

1) Paeksiard featune Eliminaton (BrE)
1) Rceursve feature Elimination( RFE)

eatue Learnind
R deafues Algrnita

st of TAbhrhrciake Larning
|L |Purdorovane
|features duture Algort

fotutes ane ranked baed on -the mod

’REt

celcorcolation Co effiuend)
ANOVA
latimCo-esficient baud

ik ud -for, caorial data)-bevd

¬- Eahehed
DiRae
M 40 60 loD

20 70 - 1o0

70 Bo

CEaD: | ooX130

(o-3s)2
35
l60-65)
65 6S

F): &0 2s)2

3S
(70-65)2
FND 6S

MD
 DOf Nh.of inde kendont voriablu
e- cotumns

of
Slur: Gtardord dewiotion

(done on g bohylaton aad on

pairad t- tesk ortai ierventon )

ehotutro diaae
192
fe- 2i D0, 108, 2|0 148 |05 yße-i
tyfe. 2.
Gobal moan, Et4+224 )
210 dyk.2.
300

DOB

Do)
 Some

m Dof 14

DOB 3- | 2

ss = (value- mean)

tes (ar0-65)(a 192-165) (i4s-18) 2

Shfe.2: (b- 43 (108- 48 (210-19
(205-224)Ý(30-229)4 ('82-214) -