ISYEl 6501l Finall Examl (Latestl 2025/l 2026l Update)l Introductionl tol Analyticsl Modelingl Reviewl |Questionsl &l Answers|l Gradel A|l 100%l Correctl (Verifiedl Solutions)
Q:l A/Bl testing
Answer:
Wheneverl wel wantl tol choosel betweenl 2l alternatives.
Asl longl asl thel followingl 3l thingsl arel true:
1st,l wel needl tol bel ablel tol collectl al lotl ofl datal quicklyl enoughl tol getl anl answerl inl timel tol usel it.2nd,l thel datal wel collectl hasl tol bel froml al representativel samplel ofl thel whole 3rd,l thel amountl ofl datal wel collectl hasl tol bel smalll comparedl tol thel totall populationl wel wantl tol usel thel answerl on.Beforel modelingl andl beforel collectingl data
Q:l (Full)l Factoriall Design
Answer:
Testl everyl combinationl ofl variablesl inl anl experimentl tol findl eachl one'sl effect,l andl interactionl effectsl onl thel outcome.
Q:l Fractionall Factoriall Design
Answer:
Al subsetl ofl combinationsl tol testl -l selectedl combinationsl givel samel resultl asl fulll factoriall designl i.el al balancedl design Beforel modelingl andl beforel collectingl data
- / 3
Q:l Whatl approachl tol takel ifl itl isl believedl thel factorsl wel canl changel arel independent?l (Factoriall design)
Answer:
Testl al subsetl ofl combinationsl andl usel regressionl tol estimatel thel effectl ofl eachl choicel Beforel modelingl andl beforel collectingl data
Q:l Explorationl vs.l Exploitation
Answer:
Explorationl -l focusingl onl gettingl morel information Exploitationl -l gettingl immediatel value
Q:l Multi-Armedl Banditl Problem
Answer:
Exploration/Exploitationl principlel Severall slotl machines,l notl knownl whichl hasl thel highestl payout,l sol mustl testl alll (K)l alternatives
1stl testl =l equall probabilityl 2ndl testl =l updatel probabilitiesl basedl offl ofl 1stl testl (wel canl alsol change:l #l ofl tests,l howl wel updatel probabilities,l changel howl wel assignl newl tests)
Q:l Whatl needsl tol bel thel casel whenl matchingl datal tol al probabilityl distributionl tol gainl insightl basedl onl howl thel distributionl isl derived?
Answer:
Thel onlyl informationl wel havel aboutl al datal pointl isl thel response,l orl whenl itl wouldl bel hardl tol collectl andl analyzel additionall information
Q:l Whatl isl thel Bernoullil distributionl isl usefull tol model?
Answer:
Al singlel event.l i.el flippingl al coin,l willl itl rainl orl not?,l willl Il getl thisl jobl offerl orl not?Onlyl reallyl usefull whenl youl putl manyl ofl theml togetherl (flipl al coinl 10,000l times)
- / 3
Q:l Describel al Bernoullil distributionl inl termsl ofl al coinl tossl test
Answer:
Probabilityl (p)l thatl al singlel coinl flipl comesl upl headsl andl probabilityl (1-p)l thatl thel coinl comesl upl tails
Q:l Definel al Binomiall distribution
Answer:
Thel probabilityl ofl gettingl xl yesl answersl outl ofl nl independentl Bernoullil trials,l eachl withl thel probabilityl p
Q:l Whenl isl thel normall distributionl usefull asl anl estimatel forl thel Binomiall distribution?
Answer:
Whenl nl isl large,l andl forl modelingl errorsl (predictivel models)
Q:l Whatl isl thel questionl tol describel al Geometricl distribution?
Answer:
Howl manyl (Bernoulli)l trialsl arel neededl beforel wel getl anl answerl ofl al certainl type?
Q:l Whatl isl thel Poissonl distributionl goodl atl modeling?
Answer:
Randoml arrivalsl ofl peoplel tol lines,l queuesl etc -l Thel functionl givesl thel probabilityl thatl xl peoplel dol arrivel givenl thel averagel arrivall ratel (lambda) -l assumesl arrivalsl arel independent,l andl identicallyl distributedl (i.i.d)
Q:l Whatl isl thel Exponentiall distributionl goodl atl modeling?
Answer:
- / 3