0% found this document useful (0 votes)
2 views12 pagesMachine Learning Assignment NO.1
This is Assignment NO.1 of Machine Learning - Semester 7 Mumbai University Syllabus.
Uploaded by
rishitestacc20Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
2 views12 pagesMachine Learning Assignment NO.1
This is Assignment NO.1 of Machine Learning - Semester 7 Mumbai University Syllabus.
Uploaded by
rishitestacc20Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
ont.
11 Sabka Malik Atma I!
Vishwatmak Jangli Maharaj Ashram Trust's
ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
A ig ment wo. 1
Expleln about machine learning tye! e
ag:
eacichia Pog_t: Spread
Toreligence Cat) thal enables System to learn.
Peco data € Poaprove ely performance On
4 task Ditheout being plicthy progt om mm
ai favalves tranining agorishm on _cleta te
make Pred chons lasstey objects or make
decisions -
types of enachine leornfa
2 . ~~ =,
\. “Supervised Jeong 2-The algorithm leorn
Prom labeled claso -te make Prdictien on Ned
Onseen dota:
2: Oncuperviced learning ihe algorthe idenh’.
fies patterns ar relat aships In Volabebed
Lert or
Rem? Qupewised learnfag ¢ The algor?snon
learn Pear a Corals wonas labled on
ULalobelee) deta -
4. Reinforceca + \earnfasy 2TTRe culgerfihns Leen —
athroysh tal £ ermmor By tattxcting Osh
An environment § receiving we10.
\Benal+ eg. MW Sabka Malik Atma It
Vishwatmak Jangli Maharaj Ashram Trust's
ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
—rssu
S Dore Qucling 2 machine learning clgoxtthma
mequrve high= quay clata +o Tears
3.
vole} Gucity can lead +o
loc Se cl Or Paaccurate models.
2. overfisting 2 When o model Ts tec complex, P+
J one >
may FP+ the trainfog cata too closdy te MAL gy
fa poor performance On ne oleata-
> Undenfitiag e When q mode) Ps too Simple,
PE mew nor capture +he Underluia
°
attern
Ta =the cletta
> °
wesulhing fa poor Perform ance
4. Bias & fairness 3
§ machine. learning models
Can perpetucite biases present fa phe
oetta leading Tt _OnPeeir Outcome
S. sraterpretauhby | fy
ty { Complex cnachia learning _
Medels Can be AiPpicul} ++ interpret , a
cocking 2h Haven tng, are Onclensstand -the
reasant
—.
Isehind heir Predicttoac.
E. Scotlab?ixy 2 machine learning models acan
be Com pwEtonaliy Patenciv Vequiria
igciPfosat TeSource -to ‘rain & depiey-— - 1 Sabka Malik Atma It
A» Z\ Vishwatmak Jang Mahara Ashram Trust's
‘keg, ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
2. | DeserSbe about Constructing Decision Tree
Using Got Tndex CPegressfen) . classtecat on
& Rave sstan Tree:
Canstructing Decision —rees Using Gin? tna
1 Starck DRh-the met pode ihe ala orthm
Storts Oth the woot plode, Which sreprese nt
sthe Ontive slat Set.
2. chooSe -the best Split t—the alqosihimn Select
che feajure § Sel?t potat that result fa
the \orgest reduction in Gini rndex.
3. Sait the data {The data 6 a Split foto
-kdo Subsets based on jhe chosen peature.
Ss We point -
Sp
4. Reorston § The algorvthes recursively copplie
the abou Yeps -to Gach Subsot Oats:
A Stopping ctentan 2s mey.
\assipfaeatfon 2 Pesression trees Ceaepr)y-
carr fs a Popular clectsfen tree alaovsthn,
sthet Oses ~+he fa? Tndes the Por lass eccad-
fon § Vaviance re, luctYon for rear, ion
rs
CARY for ClassiPoab on.
\._ Gini “Endey 3 Capt Uses the Gra? -ndex +o
Measure phat fenpuriy of a node .
2. Splitting 3 Cart spl; the dato based on th HW Sabka Malik Atma It
Vishwatmak Jangli Maharaj Ashram Trust's
ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
/
Z
fot that resud Pa _-th
t Pat lex.
For ression ~
onwance Peducton 1: CART USes _—_
mreduction to measure the weduction in
Newttance ab Esch other.
IH na | CAR iis the dato based on
eat ecotu: 2 iat Shak Wesutt
athe l\orgest recluchon Pa Voron 1 Sabka Malik Atma It
Vishwatmak Jangli Maharaj Ashram Trust's
ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
ee CINOLOGY & RE
Ss
TaveWwe
ame Class* on oipferent
or he 2 ini
fetion TS made b:
cla ssi.
3
a
lo
comipinin
er
ich on
} c 1 Sabka Malik Atma It
\Vishwatmak Jangll Maharaj Ashram Trust's
ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
:
vere a)
ecsictyion.
the avera
ve
of Ey classi
aby
Ss.
jerarchic
Wierarchical
oF
ach cle ssigs
Of classe.
ASS f 1 Sabka Malik Atma I!
Vishwatmak Jangli Maharaj Ashram Trust's
ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
Explain _cbout Svm for linear & Nen\ineor
Classi Ficetien +
upport Vector machines CSvmS)
Svms_ are a -+ype of sppemvsed leaming
wSthen Osed For class? Pycuh’en & reqression
Fasks. hey are particularly, eprective to.
high -dimensfonw) Spaces & Can handle beth
“Ss
\fnear & non-linear classiPfoutien problems.
Ineo cluss*fcadion
ta _Vnear classification Svms' alm +o Pind +e 9
hy Seplane —phek eparatec the classes_ wrth
The Maximus Merrgin. He byperplan Ps
clenfnes by a Ser of Support Vectors. hich
are Ye ingtances cleSe to the hyperplane -
L_Piyperpleane 3 A hypexplan, Te a dectsPon
boundory hot Sporad: phe clasSes -
2. Support vectors § Support vectors are the
fnstances cloSes} -}o Jhe byperplane
B+ margio §-1Ke margin Ss the d?stance
| bet Ween he byper Plane £ tne Support
Nectors -
| Whon = linear classi Peuttan
Zn _non-linea Jass'Peahon Syms Osa a
\ceene) hicks “te_meap he deta Pnte_a_hfgher:
£0. ac here the classes are. -
Vesey Separable - 1 Sabka Malik Atma I!
ki /\\
Vishwatmak Jangli Maharaj Ashram Trust's
sez; ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
Kernel rick
—MWe_kerne) feck Pavoive
S: tag Kernel
Punct?on ~to map =the dota %,
oto a higher.
climenstonal Space
P-phe dato potas
Space.
Without © piichtly Compuhin:
the Coordinates o
Pa ther
PTypes OF kesners
\. Unear kernel &
B linear kerne) Osed For
Vnear class? Routon Problems -
2. Poly nomia) Kersne) 2
3.
A_Polynomra) Kernel).
1S OSes) For non - |i a,
o CRSP) kesned § An
RBE Kernel Ps Used For non- l?near
Cla
: x_olaSS?PPeation fro blems.
3. Radia) Basic Buinctio
iRfcatfon problems.
A Y * \Vishwatmak Jangli Maharaj Ashram Trust's
2 LX, ATMA MALIK INSTITUTE OF TECHNOLOGY & RESEARCH
5. | Explain about Graph Based Clustering Clustens'no
With mratme Sponning rteee- :
Graph - Based clustering, DIth minmmeal
—Tree Cmst>
pannra
f ‘S
Graph -based clustering Using mst Ps_a -lechnigue
hak groups data potats Tato clusters based on
ely Sfm@jowHes ers hod T4 Works ¢
Step 1. Graph Construction
= Represent date pornis as vertfoes TAO Gauph -
~ Calewlate the distances bet” dlate gcints =
Wepresen} +hem as edges.
Step 2 / Mst calculaton
= Compute -the eotnimum pane Tas, spree Cmsr) of
he graph Using odgoshm Bee fevrUskal's or
Pxims
-—The MST Connects GI! verHfees WH the
mafatmum otal edge Weight, on, wins, no _cucles .
yep D>: Clustering
- Remove
