(精品)060_Techniques_of_Data_Analysis.ppt
《(精品)060_Techniques_of_Data_Analysis.ppt》由会员分享,可在线阅读,更多相关《(精品)060_Techniques_of_Data_Analysis.ppt(65页珍藏版)》请在得力文库 - 分享文档赚钱的网站上搜索。
1、TechniquesofDataAnalysisAssoc.Prof.Dr.AbdulHamidb.Hj.MarImanDirectorCentreforRealEstateStudiesFacultyofEngineeringandGeoinformationScienceUniversitiTekbnologiMalaysiaSkudai,JohorObjectiveslOverall:ReinforceyourunderstandingfromthemainlecturelSpecific:*Conceptsofdataanalysis*Somedataanalysistechnique
2、s*SometipsfordataanalysislWhatIwillnotdo:*ToteacheverybitandpiecesofstatisticalanalysistechniquesDataanalysis“TheConcept”lApproachtode-synthesizingdata,informational,and/orfactualelementstoanswerresearchquestionslMethodofputtingtogetherfactsandfigurestosolveresearchproblemlSystematicprocessofutilizi
3、ngdatatoaddressresearchquestionslBreakingdownresearchissuesthroughutilizingcontrolleddataandfactualinformationCategoriesofdataanalysislNarrative(e.g.laws,arts)lDescriptive(e.g.socialsciences)lStatistical/mathematical(pure/appliedsciences)lAudio-Optical(e.g.telecommunication)lOthersMostresearchanalys
4、es,arguably,adoptthefirstthree.Thesecondandthirdare,arguably,mostpopularinpure,applied,andsocialsciencesStatisticalMethodslSomethingtodowith“statistics”lStatistics:“meaningful”quantitiesaboutasampleofobjects,things,persons,events,phenomena,etc.lWidelyusedinsocialsciences.lSimpletocomplexissues.E.g.*
5、correlation*anova*manova*regression*econometricmodellinglTwomaincategories:*Descriptivestatistics*InferentialstatisticsDescriptivestatisticslUsesampleinformationtoexplain/makeabstractionofpopulation“phenomena”.lCommon“phenomena”:l*Association(e.g.1,2.3=0.75)l*Tendency(left-skew,right-skew)l*Causalre
6、lationship(e.g.ifX,then,Y)l*Trend,pattern,dispersion,rangelUsedinnon-parametricanalysis(e.g.chi-square,t-test,2-wayanova)Examplesof“abstraction”ofphenomenaExamplesof“abstraction”ofphenomena%prediction errorInferentialstatisticslUsingsamplestatisticstoinfersome“phenomena”ofpopulationparameterslCommon
7、“phenomena”:cause-and-effect*One-wayr/ship*Multi-directionalr/ship*RecursivelUseparametricanalysisY1=f(Y2,X,e1)Y2=f(Y1,Z,e2)Y1=f(X,e1)Y2=f(Y1,Z,e2)Y=f(X)ExamplesofrelationshipDep=9t215.8Dep=7t192.6Whichonetouse?lNatureofresearch*Descriptiveinnature?*Attemptsto“infer”,“predict”,find“cause-and-effect”
8、,“influence”,“relationship”?*Isitboth?lResearchdesign(incl.variablesinvolved).E.g.lOutputs/resultsexpected*researchissue*researchquestions*researchhypothesesAtpost-graduatelevelresearch,failuretochoosethecorrectdataanalysistechniqueisanalmostsureingredientforthesisfailure.Commonmistakesindataanalysi
9、slWrongtechniques.E.g.lInfeasibletechniques.E.g.Howtodesignex-anteeffectsofKLIA?Developmentoccurs“before”and“after”!Whatisthecontroltreatment?Furtherexplanation!lAbuseofstatistics.E.g.lSimplyexcludeatechniqueNote:NowaycanLikertscalingshow“cause-and-effect”phenomena!IssueData analysis techniquesWrong
10、 techniqueCorrect techniqueTo study factors that“influence”visitors to come to a recreation site“Effects”of KLIA on the development of SepangLikert scaling based on interviewsLikert scaling based on interviewsData tabulation based on open-ended questionnaire surveyDescriptive analysis based on ex-an
11、te post-ante experimental investigationCommonmistakes(contd.)“Abuseofstatistics”IssueData analysis techniquesExample of abuseCorrect techniqueMeasurethe“influence”ofavariableonanotherUsingpartialcorrelation(e.g.Spearmancoeff.)UsingaregressionparameterFindingthe“relationship”betweenonevariablewithano
12、therMulti-dimensionalscaling,LikertscalingSimpleregressioncoefficientToevaluatewhetheramodelfitsdatabetterthantheotherUsingR2Manya.o.t.Box-Cox2testformodelequivalenceToevaluateaccuracyof“prediction”UsingR2and/orF-valueofamodelHold-outsamplesMAPE“Compare”whetheragroupisdifferentfromanotherMulti-dimen
13、sionalscaling,LikertscalingManya.o.t.two-wayanova,2,ZtestTodeterminewhetheragroupoffactors“significantlyinfluence”theobservedphenomenonMulti-dimensionalscaling,LikertscalingManya.o.t.manova,regressionHowtoavoidmistakes-UsefultipslCrystalizetheresearchproblemoperabilityofit!lReadliteratureondataanaly
14、sistechniques.lEvaluatevarioustechniquesthatcandosimilarthingsw.r.t.toresearchproblemlKnowwhatatechniquedoesandwhatitdoesntlConsultpeople,esp.supervisorlPilot-runthedataandevaluateresultslDontdoresearch?PrinciplesofanalysislGoalofananalysis:*Toexplaincause-and-effectphenomena*Torelateresearchwithrea
15、l-worldevent*Topredict/forecastthereal-worldphenomenabasedonresearch*Findinganswerstoaparticularproblem*Makingconclusionsaboutreal-worldeventbasedontheproblem*LearningalessonfromtheproblemlDatacant“talk”lAnanalysiscontainssomeaspectsofscientificreasoning/argument:*Define*Interpret*Evaluate*Illustrat
16、e*Discuss*Explain*Clarify*Compare*ContrastPrinciplesofanalysis(contd.)Principlesofanalysis(contd.)lAnanalysismusthavefourelements:*Data/information(what)*Scientificreasoning/argument(what?who?where?how?whathappens?)*Finding(whatresults?)*Lesson/conclusion(sowhat?sohow?therefore,)lExamplePrinciplesof
17、dataanalysislBasicguidetodataanalysis:*“Analyse”NOT“narrate”*Gobacktoresearchflowchart*Breakdownintoresearchobjectivesandresearchquestions*Identifyphenomenatobeinvestigated*Visualisethe“expected”answers*Validatetheanswerswithdata*DonttellsomethingnotsupportedbydataPrinciplesofdataanalysis(contd.)Sho
18、ppersNumberMaleOldYoung64FemaleOldYoung1015MorefemaleshoppersthanmaleshoppersMoreyoungfemaleshoppersthanyoungmaleshoppersYoungmaleshoppersarenotinterestedtoshopattheshoppingcomplexDataanalysis(contd.)lWhenanalysing:*Beobjective*Accurate*TruelSeparatefactsandopinionlAvoid“wrong”reasoning/argument.E.g
19、.mistakesininterpretation.Introductory Statistics for Social SciencesBasic conceptsBasic conceptsCentral tendencyCentral tendencyVariabilityVariabilityProbabilityProbabilityStatistical Statistical ModellingModellingBasicConceptslPopulation:thewholesetofa“universe”lSample:asub-setofapopulationlParame
20、ter:anunknown“fixed”valueofpopulationcharacteristiclStatistic:aknown/calculablevalueofsamplecharacteristicrepresentingthatofthepopulation.E.g.=meanofpopulation,=meanofsampleQ:WhatisthemeanpriceofhousesinJ.B.?A:RM210,000J.B.houses=?SSTDSTSD1=300,000=120,0002=210,0003BasicConcepts(contd.)lRandomness:M
21、anythingsoccurbypurechancesrainfall,disease,birth,death,.lVariability:Stochasticprocessesbringinthemvariousdifferentdimensions,characteristics,properties,features,etc.,inthepopulationlStatisticalanalysismethodshavebeendevelopedtodealwiththeseverynatureofrealworld.“CentralTendency”MeasureAdvantagesDi
22、sadvantagesMean(Sumofallvaluesno.ofvalues)BestknownaverageExactlycalculableMakeuseofalldataUsefulforstatisticalanalysisAffectedbyextremevaluesCanbeabsurdfordiscretedata(e.g.Familysize=4.5person)CannotbeobtainedgraphicallyMedian(middlevalue)NotinfluencedbyextremevaluesObtainableevenifdatadistribution
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- 精品 060 _Techniques_of_Data_Analysis
限制150内