ISYEl 6501l Midterml Examl 1l (Latestl 2025/l 2026l Update)l Introductionl tol Analyticsl Modelingl Guidel |Questionsl &l Answers|l Gradel A|l 100%l Correctl (Verifiedl Solutions)
Q:l Al surveyl ofl 25l peoplel recordedl eachl person'sl familyl sizel andl typel ofl car.l Whichl ofl thesel isl al datal point?A.l Thel 14thl person'sl familyl sizel andl carl type B.l Thel 14thl person'sl familyl size C.l Thel carl typel ofl eachl person
Answer:
A.l Al datal pointl isl alll thel informationl aboutl onel observation
Q:l Thel fartherl thel wronglyl classifiedl pointl isl froml thel linel ___
Answer:
Thel biggerl thel mistakel we'vel made
Q:l Thel terml includingl thel marginl getsl largerl sol thel importancel ofl al largel marginl outl weightsl avoidingl mistakesl andl classifyingl knownl datal samples.
Answer:
Asl lambdal getsl larger
Q:l Thatl terml alsol dropsl towardsl zero,l sol thel importancel ofl minimizingl mistakesl andl classifyingl knownl datal pointsl outweighsl havingl al largel margin.
Answer:
Asl lambdal dropsl towardsl zero 1 / 4
Q:l Whatl canl SVMsl bel usedl for
Answer:
tol findl al classifierl withl maximuml seperationl orl marginl betweenl thel twol setsl ofl points?
Q:l Whenl tol usel SVM?
Answer:
Ifl it'sl impossiblel tol avoidl classificationl errors,l SVMl canl findl al classifierl thatl tradesl offl reducingl errorsl andl enlargingl thel margin.
Q:l Errorl forl datal pointl j
Answer:
Whatl doesl thisl formulal describe?
Q:l Totall error
Answer:
Whatl doesl thisl formulal describel ?
Q:l Tol maximizel thel distancel betweenl thel twol linesl whatl dol wel needl tol minimize?
Answer:
Q:l m_jl >l 1
Answer:
Whatl valuel dol wel givel forl morel costlyl errors
Q:l Givingl al badl loanl isl twicel asl costlyl asl withholdingl al goodl loan? 2 / 4
Answer:
Whatl doesl thisl meanl inl thel contextl ofl givingl al loan?
Q:l m_jl
Answer:
Whatl valuel dol wel givel forl lessl costlyl errors?
Q:l Whyl isl itl importantl tol scalel ourl datal whenl usingl SVM?
Answer:
We'rel lookingl tol minimizel thel suml ofl thel squaresl ofl thel coefficients,l butl ifl ourl datal hasl veryl differentl scalesl al smalll changel inl onel couldl swampl al hugel changel inl thel other.
Q:l whatl doesl itl signifyl whenl al coefficientl forl al classifierl isl closel tol zero
Answer:
itl meansl thel correspondingl attributel isl probablyl notl relevant
Q:l Whatl dol kernell methodsl allowl forl inl SVMs
Answer:
nonlinearl classifiers
Q:l Whatl isl thel commonl rangel forl scaledl data?
Answer:
betweenl 0l andl 1
Q:l Whatl isl thel formulal forl min-maxl scaling?
Answer: 3 / 4
findl minl andl maxl forl al factor
Q:l whatl isl commonl standardizationl andl itsl formula?
Answer:
scalingl tol al normall distributionl withl al meanl ofl 0l andl standardl deviationl ofl 1.
Q:l whatl isl thel formulal forl generall scalingl betweenl bl andl a
Answer:
Q:l Whenl dol youl usel scaling?
Answer:
Datal inl al boundedl rangel (e.g.,l neurall networks,l RGBl values,l SATl scores,l battingl averages)
Q:l Whenl dol youl usel standardization?
Answer:
PCAl orl clustering
Q:l Whenl isl KNNl used?
Answer:
Usedl forl solvingl classificationl problemsl inl whichl therel arel morel thanl twol classes.
Q:l Howl dol youl deall withl attributesl thatl mightl bel morel importantl thanl othersl inl KNN?
Answer:
Youl weightl eachl dimension'sl distancel different.l Thel largerl thel weightl thel higherl thel impact.
- / 4
Answer:
Whatl valuel dol wel givel forl lessl costlyl errors?
Q:l Whyl isl itl importantl tol scalel ourl datal whenl usingl SVM?
Answer:
We'rel lookingl tol minimizel thel suml ofl thel squaresl ofl thel coefficients,l butl ifl ourl datal hasl veryl differentl scalesl al smalll changel inl onel couldl swampl al hugel changel inl thel other.
Q:l whatl doesl itl signifyl whenl al coefficientl forl al classifierl isl closel tol zero
Answer:
itl meansl thel correspondingl attributel isl probablyl notl relevant
Q:l Whatl dol kernell methodsl allowl forl inl SVMs
Answer:
nonlinearl classifiers
Q:l Whatl isl thel commonl rangel forl scaledl data?
Answer:
betweenl 0l andl 1
Q:l Whatl isl thel formulal forl min-maxl scaling?
Answer: 3 / 4
findl minl andl maxl forl al factor
Q:l whatl isl commonl standardizationl andl itsl formula?
Answer:
scalingl tol al normall distributionl withl al meanl ofl 0l andl standardl deviationl ofl 1.
Q:l whatl isl thel formulal forl generall scalingl betweenl bl andl a
Answer:
Q:l Whenl dol youl usel scaling?
Answer:
Datal inl al boundedl rangel (e.g.,l neurall networks,l RGBl values,l SATl scores,l battingl averages)
Q:l Whenl dol youl usel standardization?
Answer:
PCAl orl clustering
Q:l Whenl isl KNNl used?
Answer:
Usedl forl solvingl classificationl problemsl inl whichl therel arel morel thanl twol classes.
Q:l Howl dol youl deall withl attributesl thatl mightl bel morel importantl thanl othersl inl KNN?
Answer:
Youl weightl eachl dimension'sl distancel different.l Thel largerl thel weightl thel higherl thel impact.
- / 4