應(yīng)用商務(wù)統(tǒng)計學(xué)講義-第一章-中英文對照課件

Ch0:FirstThingsFirst

and

Ch1:Definingandcollectingdata

LearningObjectivesWhatisStatistics?什么是統(tǒng)計學(xué)BasicconceptsinStatistics統(tǒng)計學(xué)中的基本概念Data,variable,population,sample,parameter,statistic,etc…數(shù)據(jù)、變量、人口樣本統(tǒng)計量、參數(shù)，等等……Data/variabletypes數(shù)據(jù)/變量類型Howtocollectdata如何收集數(shù)據(jù)Thedifferentwaystocollectasample收集樣本的不同方法Thetypesofsurveyerrors調(diào)查誤差的類型1Ch0:FirstThingsFirst

and

WhatisStatistics?Statisticsreferstomethodsthathelptransformdataintousefulinformationfordecisionmakers.統(tǒng)計指的是幫助決策者將數(shù)據(jù)轉(zhuǎn)化為有用信息的方法。Statisticsisawayofthinkingthatcanleadtobetterdecisions.統(tǒng)計是一種可以帶來更好決策的思維方式。2WhatisStatistics?StatisticsWhyStatistics?Intoday’sdigitalworldeverincreasingamountsofdataaregathered,stored,reportedon,andavailableforfurtherstudy.Businessinformationsystems?在當今的數(shù)字世界中，越來越多的數(shù)據(jù)被收集、存儲、報告，并可供進一步研究。-商業(yè)信息系統(tǒng)Youheartheworddataeverywhere.你到處都聽到“數(shù)據(jù)”這個詞。Dataarefactsabouttheworldandareconstantlyreportedbyaneverincreasingnumberofsources.數(shù)據(jù)是關(guān)于世界的事實，并且不斷地被越來越多的來源所報道。3WhyStatistics?Intoday’sdigiToProperlyApplyStatisticsYouShouldFollowAFrameworkToMinimizePossibleErrors

為了正確地應(yīng)用統(tǒng)計數(shù)據(jù)，您應(yīng)該遵循一個框架，以盡量減少可能出現(xiàn)的錯誤。

InthiscoursewewilluseDCOVADefinethedatayouwanttostudyinordertosolveaproblemormeetanobjectiveCollectthedatafromappropriatesourcesOrganizethedatacollectedbydevelopingtablesVisualizethedatabydevelopingchartsAnalyzethedatacollectedtoreachconclusionsandpresentresults在這個過程中我們將使用DCOVA-定義你想研究的數(shù)據(jù)，以解決問題或達到一個目標。-從適當?shù)膩碓词占瘮?shù)據(jù)-組織開發(fā)表收集的數(shù)據(jù)-通過開發(fā)圖表來可視化數(shù)據(jù)-分析收集到的數(shù)據(jù)，得出結(jié)論并給出結(jié)果4ToProperlyApplyStatisticsYUsingTheDCOVAFrameworkHelpsYouToApplyStatisticsTo:

使用DCOVA框架幫助你申請統(tǒng)計：

Summarize&visualizebusinessdata總結(jié)和可視化業(yè)務(wù)數(shù)據(jù)Reachconclusionsfromthosedata從這些數(shù)據(jù)中得出結(jié)論Makereliableforecastsaboutbusinessactivities對業(yè)務(wù)活動作出可靠的預(yù)測Improvebusinessprocesses改進業(yè)務(wù)流程5UsingTheDCOVAFrameworkHelpBusinessAnalytics:TheChangingFaceOfStatistics

商業(yè)分析：統(tǒng)計數(shù)據(jù)的變化Useinformationsystemsmethodstocollectandprocessdatasetsofallsizes,includingverylargedatasetsthatwouldotherwisebehardtoexamineefficiently.?使用信息系統(tǒng)方法收集和處理各種大小的數(shù)據(jù)集，包括非常大的數(shù)據(jù)集，否則很難有效地檢查這些數(shù)據(jù)集。Usestatisticalmethodstoanalyzeandexploredatatouncoverunforeseenrelationships.

?使用統(tǒng)計方法分析和探索數(shù)據(jù)，以發(fā)現(xiàn)不可預(yù)見的關(guān)系。Usemanagementsciencemethodstodevelopoptimizationmodelsthatimpactanorganization’sstrategy,planning,andoperations.

?使用管理科學(xué)方法開發(fā)影響組織戰(zhàn)略、規(guī)劃和運作的優(yōu)化模型。Thegrowthof“BigData”spurstheuseofbusinessanalytics?“大數(shù)據(jù)”的增長刺激了商業(yè)分析的應(yīng)用“Bigdata”orverylargedatasetsarearisingbecauseoftheautomaticcollectionofhighvolumesofdataatveryfastrates.

?“大數(shù)據(jù)”或非常大的數(shù)據(jù)集的出現(xiàn)，是因為以非?？斓乃俾首詣邮占罅繑?shù)據(jù)。6BusinessAnalytics:TheChangiDataVocabulary數(shù)據(jù)的詞匯Data:measurementsthatarecollected,recorded,andsummarizedforpresentation,analysis,andinterpretation–數(shù)據(jù)：收集、記錄和總結(jié)用于陳述、分析和解釋的測量Variable:characteristicoftheelementswhosevaluesmaydifferfromelementtoelementandisofinteresttothedatacollector變量：元素的特征，其值可能不同于元素到元素，并且對數(shù)據(jù)收集器感興趣。Element:anentityorobjectonwhichdataarecollected.Alsocalledcase,subject,individual,item-元素：收集數(shù)據(jù)的實體或?qū)ο蟆Ｒ卜Q案件、主體、個人、項目Observation:measurementofavariableonasingleelement-觀察：單個元素上變量的測量7DataVocabulary數(shù)據(jù)的詞匯Data:measDataVocabularyCaseNameAgeIncomePositionGender1Frieda45$67,100PersonneldirectorF2Stefan3256,500OperationsmanagerM3Barbara5588,200MarketingVPF4Donna2759,000StatisticianM5Larry4636,000SecurityguardF6Alicia5268,500ComptrollerM7Alex6592,500ChiefexecutiveM8Jaime5071,200PublicrelationsF5variables8subjects/elements/individuals/items40observations8DataVocabularyCaseNameAgeIncoDataVocabularyTypesofVariables變量類型Qualitative:labelsornamesforacharacteristic(position,gender,name)

-定性：特征的標簽或名稱（位置，性別，名字）Quantitative:measurementofamountorquantity-定量：量或量的測量Discrete(counting)(#offamilynumbers):limitedvaluesinarange離散（計數(shù)）（#家屬）：在一個有限的范圍值Continuousvariable(measuring)(age,income):anyvalueinarange?連續(xù)變量（測量）（年齡，收入）：某一范圍內(nèi)的任何值9DataVocabularyTypesofVariabDataVocabularyVariabletypesQualitative(Nominal,categorical)Quantitative(Numerical)DiscreteContinuousWords?Integers?10定性的（名義的，明確的變量類型定量（數(shù)值）分離的，不相關(guān)聯(lián)的連續(xù)的整數(shù)？語言?DataVocabularyVariabletypesQAmountofInformation1.Nominallevel2.Ordinallevel3.Intervallevel4.RatiolevelNoorderordered/rankede.g.EyecolorRatingofaprofessor

AbsolutezeroDifferenceismeaningfulRatioisalsomeaningfulSalaryLevels/ScalesofmeasurementNotruezeroDifferenceismeaningfulRatioisnotmeaningfulTemperatureFourLevelsofMeasurement11測量的四個層次信息量水平/測量尺度1。標稱等級2。順序?qū)哟?.區(qū)間水平4.率水平AmountofInfoFourLevelsofMeasurementQualitativedata:NominalandOrdinallevelsNominalscale/level:Valuesrepresentcategoryorgroupmembershipofelements.Onlyshowdifference).Noorderimplied.?定性數(shù)據(jù)：名詞和序數(shù)級-名義量表/級別：值表示元素的類別或組成員關(guān)系。僅表現(xiàn)出差異）。無訂單暗示。Ordinalscale/level:valuesconveylessthan,equalto,andgreaterthanrelationshipsamongelements,i.e.therelativeranksoftheelementswithrespecttotheirvaluesforthevariableinquestion(onebetterthananother?)(ratingsofcustomerservice:good,average,poor)–-序數(shù)量/等級：值傳遞小于，等于，大于元素之間的關(guān)系，即相對于變量的值的元素的相對秩（一個比另一個更好？）（客戶服務(wù)等級：好的，一般的，差的）

12FourLevelsofMeasurementQualFourLevelsofMeasurementQuantitativedata:IntervalandRatioScalesIntervalscale/level:thedifferencebetweenmeasurementsisameaningfulquantitybutdoesnotinvolveatruezeropointFahrenheittemperature:differencebetween68-70isthesameas70-72.0degreedoesnotmeannotemperature.?定量數(shù)據(jù)：區(qū)間和比率標度-間隔刻度/水平：測量之間的差異是有意義的數(shù)量，但不包括真正的零點。?華氏溫度之間的差別是：6870-72相同。0度并不意味著沒有溫度。Ratioscale:valuescantakeonanaturalorabsolutezeroandratioismeaningfulSalary:0meansnoincome.40000istwiceasmuchas20000.80000istwiceasmuchas40000.比例標尺：值可以是自然的或絕對的零，比率是有意義的。?工資：0表示沒有收入。40000是20000的兩倍。80000是40000的兩倍。13FourLevelsofMeasurementQuanDatacollectionmethods

數(shù)據(jù)收集方法Datadistributedbyanorganizationoranindividual由組織或個人分發(fā)的數(shù)據(jù)Adesignedexperiment設(shè)計的實驗Asurvey/Anobservationalstudy調(diào)查/觀察性研究Datacollectedbyongoingbusinessactivities正在進行的業(yè)務(wù)活動收集的數(shù)據(jù)14Datacollectionmethods

數(shù)據(jù)收集方法SourcesofData數(shù)據(jù)來源PrimarySources:Thedatacollectoristheoneusingthedataforanalysis§主要來源：數(shù)據(jù)采集器是一個用數(shù)據(jù)分析Datafromapoliticalsurvey§從政治的調(diào)查數(shù)據(jù)Datacollectedfromanexperiment§試驗數(shù)據(jù)采集Observeddata§觀測數(shù)據(jù)SecondarySources:Thepersonperformingdataanalysisisnotthedatacollector次要來源：進行數(shù)據(jù)分析的人不是數(shù)據(jù)采集器Analyzingcensusdata分析人口普查數(shù)據(jù)Examiningdatafromprintjournalsordatapublishedontheinternet.檢查數(shù)據(jù)從印刷刊物或在互聯(lián)網(wǎng)上公布的數(shù)據(jù)15SourcesofData數(shù)據(jù)來源PrimarySouMoreDefinitions更多的定義Population:entiresetofobjectsofinterest人口：一整套感興趣的對象Sample:apartofthepopulationofinterest樣本：感興趣人群中的一部分Parameter(populationcharacteristics)vs.

SampleStatistic(samplecharacteristics)參數(shù)（人口特征）與樣本統(tǒng)計（樣本特征）Inpractice,weusuallycollectasampletostudythecharacteristicsofapopulation在實踐中，我們通常收集樣本來研究人口的特征。16MoreDefinitions更多的定義PopulatioProcessofStatisticalInference/inferentialStats

統(tǒng)計推斷/推斷統(tǒng)計過程1.Populationconsistsofallelementsofinterests2.Asampleofitemsistakenandexamined4.Thestatisticisusedasanestimateofapopulationcharacteristic3.ThesampledataprovidesadescriptivestatisticAstudyshows…研究表明.17人口包括所有的利益因素。一個項目的樣本被接受和檢查。統(tǒng)計被用作人口特征的估計示例數(shù)據(jù)提供描述性統(tǒng)計。ProcessofStatisticalInferenSamplingExpenseSpeedDestructivenatureofdatacollectionInaccessibilityofsomeelements…Random/ProbabilitySamplingMethodsOtherapproachesWhysample?Howtosample?18抽樣為什么抽樣？怎么抽樣–費用–速度-數(shù)據(jù)收集的破壞性–交通不便的一些元素-隨機/概率抽樣方法–其他方法SamplingExpenseWhysample?18抽樣Howtosample:ProbabilitySampleInaprobabilitysample,itemsinthesamplearechosenonthebasisofknownprobabilities.ProbabilitySamplesSimpleRandomSystematicStratifiedCluster19如何抽樣：概率抽樣?在概率抽樣中，根據(jù)已知概率選擇樣本中的項。概率抽樣簡單的隨機有系統(tǒng)的，有規(guī)則的集群分層Howtosample:ProbabilitySamSimpleRandomSampleNumbereachunitfrom1toN?每個單元從1到nUsearandomnumbergeneratortoselectndistinctnumbersbetween1andN,inclusivelyAvailabletoolsEasiertoperformforsmallpopulationsCumbersomeforlargepopulationsRandomnumbergeneratorTableofrandomnumbersExcelfunctions:Randbetween(min,max)Add-in:SamplingWithreplacementorwithoutrep..20簡單隨機樣本?使用隨機數(shù)發(fā)生器選擇n個不同的數(shù)字之間的1和N，含可用的工具隨機數(shù)發(fā)生器?隨機數(shù)表?Excel函數(shù)：?randbetween（min，max）?加入：抽樣?更換或不帶代表對小群體來說更容易執(zhí)行?龐大人口的累贅SimpleRandomSampleNumbereacSystematicSamplingPopulationelementsareanorderedsequenceFirstsampleelementisselectedrandomlyfromthefirstkpopulationelementsThen,sampleelementsareselectedataconstantinterval,k,fromtheorderedsequenceframek=Nn

,

where:n=samplesizeN=populationsizek=sizeofselectioninterval21系統(tǒng)抽樣?人口要素是一個有序序列第一個樣本元素是從第一k個人口元素中隨機選取的。然后，樣品的元素是在一個恒定的間隔，K的選擇，從有序的序列幀試樣量，樣本大小群體大小選擇間隔的大小SystematicSamplingPopulationSystematicSamplingInvoice0118Invoice0220Invoice0221Invoice0302Invoice0308Invoice0306Invoice0402Invoice0412Invoice0514Invoice0513N=20n=5k=N/n=4First,findarandomstartingpointThenitemsfromevery4invoicesInvoice0618Invoice0620Invoice0721Invoice0802Invoice0808Invoice0906Invoice1002Invoice1012Invoice1014Invoice111322系統(tǒng)抽樣發(fā)票首先，找到一個隨機出發(fā)點每4張發(fā)票的項目SystematicSamplingInvoice011StratifiedRandomSampleWhatifourpopulationcanbeclearlydividedintosubgroupsbasedonsomecharacteristicsandwewantoursampletoincludeallsubgroups?PopulationisdividedintononoverlappingsubpopulationscalledstrataArandomsampleisselectedfromeachstratumPotentialforreducingsamplingerrorProportionate-thepercentageofthesampletakenfromeachstratumisproportionatetothepercentagethateachstratumiswithinthepopulation23分層隨機樣本?如果我們的人口可以根據(jù)一些特征明確地劃分為亞組，我們希望我們的樣本包括所有子組，該怎么辦？種群分成互不重疊的亞群，稱為地層從每個階層挑選一個隨機樣本。?減少抽樣誤差的可能性?比例-從每個階層抽取的樣本比例與每個階層在人口中所占的百分比成比例StratifiedRandomSampleWhatiStratifiedRandomSampleSupposewewanttostudytheadvertisingexpendituresforthe352largestcompaniesintheUnitedStates.?假設(shè)我們想研究美國352家最大公司的廣告支出。Tomakesurethatasampleof50companiesisafairrepresentationofthe352companies,thecompaniesaregroupedonpercentreturnonequityandasampleproportionaltotherelativesizeofthegroupisrandomlyselected.24分層隨機樣本為了確保50家公司的樣本是這352家公司的公平代表，公司按股本回報率分組，而與該集團相對規(guī)模成比例的樣本是隨機的。StratifiedRandomSampleSuppoClusterSampling(two-stagesampling)Populationisdividedintonon-overlappingclustersorareas?人口分為不重疊的群體或地區(qū)。Eachclusterisaminiature,ormicrocosm,ofthepopulationAsubsetoftheclustersisselectedrandomlyforthepopulationSimplerandomsamplingfromeachclusterselected.25（兩階段抽樣）分群[組]抽象法?每一個集群都是人口的縮影或縮影。?集群的一個子集是隨機選擇的。選擇每個群集的簡單隨機抽樣。ClusterSampling(two-stagesaNonprobabilitySampleInanonprobabilitysample,itemsincludedarechosenwithoutregardtotheirprobabilityofoccurrence.Inconveniencesampling,itemsareselectedbasedonlyonthefactthattheyareeasy,inexpensive,orconvenienttosample.Inajudgmentsample,yougettheopinionsofpre-selectedexpertsinthesubjectmatter.

Snowballsampling..26非概率抽樣在非概率抽樣，項目包括有選擇不考慮其發(fā)生的概率。-在便利抽樣中，僅根據(jù)容易、便宜或便于取樣的事實選擇項目。在一個判斷樣本中，你可以得到預(yù)先選定的專家在主題問題上的意見。滾雪球抽樣NonprobabilitySampleInanonpTypesofSurveyErrorsCoverageerrororselectionbias覆蓋誤差或選擇偏移ExistsifsomegroupsarenotincludedinthepoolandhavenochanceofbeingselectedNonresponseerrororbias無回答誤差或偏差PeoplewhodonotrespondmaybedifferentfromthosewhodorespondSamplingerrorVariationfromsampletosamplewillalwaysexistMeasurementerrorDuetoweaknessesinquestiondesign,respondenterror,andinterviewer’seffectsontherespondent27測量誤差類型如果某些組不包含在池中，并且沒有被選中的機會，則存在沒有回應(yīng)的人可能不同于那些做出反應(yīng)的人抽樣誤差從樣品到樣品的變化總是存在的?測量誤差由于問題設(shè)計的弱點，回答錯誤，以及面試官對被告的影響。TypesofSurveyErrorsCoverageDescriptivestatisticsvsInferentialStatisticsDescriptivestatistics描述統(tǒng)計（學(xué)）Tabular,graphical,andnumericalmethodsusedtosummarizeoneormorecharacteristicsofasetofdata.用于概括一組數(shù)據(jù)的一個或多個特征的表格、圖形和數(shù)值方法。Transformdatainto