《(资料收集与分析).ppt》由会员分享，可在线阅读，更多相关《(资料收集与分析).ppt（50页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、Data Collection and Analysis(資料收集與分析)By C.L.HsiehDepartment of Industrial Management Aletheia UniversityIntroduction(介紹)n n“You can observe a lot just by watching”“You can observe a lot just by watching”(你可你可以只以視覺方式來觀察許多數據以只以視覺方式來觀察許多數據)n nData gathering results a conceptual model of Data gathering
2、results a conceptual model of how the system operated how the system operated(資料收集可以產生一資料收集可以產生一個觀念式模型來解釋系統是如何運作的個觀念式模型來解釋系統是如何運作的)n nData gathering should avoid ending up with lots Data gathering should avoid ending up with lots of data but with very little useful information of data but with very
3、little useful information(資料收集應避免最後留下一堆資料但無太多有資料收集應避免最後留下一堆資料但無太多有用的資訊用的資訊)2Data Collection and AnalysisQuestions for Data Gathering(資料收集的可能問題)n nWhat is the best procedure to follow?What is the best procedure to follow?(資料收集的最佳程序為何資料收集的最佳程序為何?)?)n nWhat types of data should be gathered?What types o
4、f data should be gathered?(哪些資料種類應被收集哪些資料種類應被收集?)?)n nWhat sources should be used?What sources should be used?(資料的來源為何資料的來源為何?)?)n nWhat types of analyses should be performed on the What types of analyses should be performed on the data?data?(資料應進行何種分析資料應進行何種分析?)?)n nHow do you select the right prob
5、ability distribution How do you select the right probability distribution representing the data?representing the data?(如何找出代表資料的分配如何找出代表資料的分配?)?)n nHow should data be documented?How should data be documented?(資料如何文件化資料如何文件化?)?)3Data Collection and AnalysisGuidelines for Data Gathering(資料收集應注意事項)n nI
6、dentify triggering events:(了解啟動活動原因了解啟動活動原因)identify the causes or conditions that trigger identify the causes or conditions that trigger the activitiesthe activities，e.g.the causes of downtime:e.g.the causes of downtime:failure,idle,unavailability of stockfailure,idle,unavailability of stock (了解啟動活
7、動的原因或狀況，如停機原因：機了解啟動活動的原因或狀況，如停機原因：機器故障、閒置、缺貨器故障、閒置、缺貨)n nLook for common grouping (分群以化簡資料分群以化簡資料)the solution is to reduce the data to common the solution is to reduce the data to common behaviors and patterns behaviors and patterns (化簡資料至一般行為與樣式化簡資料至一般行為與樣式)Identify general categories Identify gen
8、eral categories(確定一般性分類確定一般性分類)4Data Collection and AnalysisGuidelines for Data Gathering(資料收集應注意事項)n nFocus on key impact factors Focus on key impact factors(處理主要影響因子處理主要影響因子)Avoid little impact information(e.g.off-hour Avoid little impact information(e.g.off-hour performance,extremely rare downtim
9、e,negligible performance,extremely rare downtime,negligible move time.)move time.)(避免影響性小的因素，如加班特例、罕見的機器故障、避免影響性小的因素，如加班特例、罕見的機器故障、可忽略的移動時間可忽略的移動時間.).)n nSeparate input variables from response variables Separate input variables from response variables(區分輸區分輸入變數與回應變數入變數與回應變數)Input variables define ho
10、w the system worksInput variables define how the system works (輸入輸入變數決定系統運作變數決定系統運作)Response variables do not“drive”model behaviorResponse variables do not“drive”model behavior5Data Collection and AnalysisGuidelines for Data Gathering(資料收集應注意事項)n nFocus on essence rather than substanceFocus on essen
11、ce rather than substanceCapture cause-effect relationships and ignore Capture cause-effect relationships and ignore meaningless details meaningless details(著重因果關係，忽略細節著重因果關係，忽略細節)Focus on the activity of using resources or the Focus on the activity of using resources or the delay of entity flow(syst
12、em abstraction)delay of entity flow(system abstraction)(重視系統抽象層面重視系統抽象層面)n nIsolate actual activity times.Isolate actual activity times.(確定真正活動時間確定真正活動時間)Exclude any extra time waiting Exclude any extra time waiting (排除額外等候時間排除額外等候時間)6Data Collection and AnalysisSteps to Gathering Data(資料收集的步驟)n nDe
13、termine data requirements(決定資料需求)n nIdentify data resources(確定資料來源)n nCollect the data(收集資料)n nMake assumptions (建立假設)n nAnalyze the data(收集資料)n nDocument and approve the data(資料文件化與驗證)7Data Collection and AnalysisDetermining Data Requirements(決定資料需求)n nStructural data(結構型資料)All the objects in the s
14、ystem to be modeledAll the objects in the system to be modeled (系統中被建模的所有物件系統中被建模的所有物件)Describe the layout of the system Describe the layout of the system (結構型資料描述系統的佈置情形結構型資料描述系統的佈置情形)Identify the items to be processed(e.g.entities,Identify the items to be processed(e.g.entities,resources,locations
15、resources,locations.).)(結構型資料確定被處理的項目，如實體、資源、結構型資料確定被處理的項目，如實體、資源、工作站等工作站等)8Data Collection and AnalysisDetermining Data Requirements(決定資料需求)n nOperational Data(作業型資料)Explain how the system operatesExplain how the system operates (解釋系統如何運作解釋系統如何運作)When,where and how events&activities take When,where
16、 and how events&activities take placeplace (解釋事件與活動發生的方式、地點與時間解釋事件與活動發生的方式、地點與時間)Consist of the logic information about the system,Consist of the logic information about the system,e.g.routing,schedules,downtime behavior and e.g.routing,schedules,downtime behavior and resource allocation.resource al
17、location.(說明系統中的運作邏輯、說明系統中的運作邏輯、如路線、排程方式、故障方式、資源分派方式如路線、排程方式、故障方式、資源分派方式)9Data Collection and AnalysisDetermining Data Requirements(決定資料需求)n nNumerical Data Numerical Data(數值型資料數值型資料)Provide quantitative information of the systemProvide quantitative information of the system (提供系統的數量資料提供系統的數量資料)Some
18、 are easy to get but some are notSome are easy to get but some are not (有些容易獲得，但有些並不容易獲得有些容易獲得，但有些並不容易獲得)e.g.capacities,arrival rates,activity timee.g.capacities,arrival rates,activity time (如工作站容量、到達率、活動時間等如工作站容量、到達率、活動時間等 )10Data Collection and AnalysisDetermining Data Requirements(決定資料需求)n nUse o
19、f a Questionnaire(sample see p.103)Use of a Questionnaire(sample see p.103)(使用問卷，樣本請見第使用問卷，樣本請見第103103頁頁)Questionnaire help gathering right informationQuestionnaire help gathering right information (問卷可以幫忙獲得正確資料問卷可以幫忙獲得正確資料)If sample data are not available,it is useful to get If sample data are not
20、available,it is useful to get at least estimate of the at least estimate of the minimumminimum,most likelymost likely,and,and maximum valuemaximum value until more precise data obtained until more precise data obtained.(若樣本資料無法獲得，至少要得到最差、最可若樣本資料無法獲得，至少要得到最差、最可能、最佳等三類估計值直到較佳資料獲得為止能、最佳等三類估計值直到較佳資料獲得為止
21、)11Data Collection and AnalysisIdentifying Data Sources(確定資料來源)n nGood sources of data Good sources of data(好的資料來源好的資料來源)Historical Records Historical Records(歷史資料，如生產量，銷售量歷史資料，如生產量，銷售量)System DocumentationSystem Documentation (系統文件，如生產計劃，設施規劃系統文件，如生產計劃，設施規劃.).)Personal Observation Personal Observat
22、ion (個人觀察，如工作取樣個人觀察，如工作取樣,時間動作研究時間動作研究.).)Personal Interviews Personal Interviews(訪談，如作業方法訪談，如作業方法,修理程序修理程序,排程排程.).)Comparison with similar systems Comparison with similar systems(與相似系統比較與相似系統比較)Vendor claim Vendor claim(零售商意見，如處理時間，新機器可靠度零售商意見，如處理時間，新機器可靠度.).)Design estimation Design estimation(設計過
23、程之估計值，處理時間設計過程之估計值，處理時間,搬運時搬運時間間.).)Research literature Research literature(文獻探討文獻探討.).)12Data Collection and AnalysisCollecting the Data(資料收集)n nDefining Entity Flow Defining Entity Flow(定義實體流定義實體流)Entity flow establishes a skeletal framework for additional Entity flow establishes a skeletal framew
24、ork for additional data be attacheddata be attached (實體流可以建立大綱式的架構實體流可以建立大綱式的架構)Follow the entity movement Follow the entity movement (實體流按實體移動路線定義實體流按實體移動路線定義)Use Entity flow diagram(EFD)Use Entity flow diagram(EFD)(使用實體流程圖使用實體流程圖)Difference between Entity flow diagram&Process Difference between En
25、tity flow diagram&Process Flowchart(Flowchart(程序流程圖程序流程圖)13Data Collection and AnalysisCollecting the Data(資料收集)Difference between Entity flow diagram&Process Difference between Entity flow diagram&Process Flowchart Flowchart(實體流程圖與程序流程圖之區別實體流程圖與程序流程圖之區別)Process Flow Process Flow chartchart(程序流程圖程序流
26、程圖)Show logical sequence of activitiesShow logical sequence of activities(顯示活動的邏輯順序顯示活動的邏輯順序)Define what it happensDefine what it happens(定義發生什麼事定義發生什麼事)Entity flow Entity flow diagram diagram(實體流程圖實體流程圖 Show physical movement of entitiesShow physical movement of entities (定義實體實際移動方式定義實體實際移動方式)Defin
27、e where it happensDefine where it happens(定義事情發生的地點定義事情發生的地點)14Data Collection and AnalysisDeveloping A Description of Operation (作業流程描述)n nDescription of Operation(Description of Operation(作業流程描述作業流程描述)Explain how entities are processed&provides the details of the EFDExplain how entities are proces
28、sed&provides the details of the EFD (解釋實體如何處理並提供解釋實體如何處理並提供EDFEDF細節細節)n nRequirements(Requirements(需求項目需求項目)Time&resource requirements of the activity or operationTime&resource requirements of the activity or operation (活動或作業的時間或資源需求活動或作業的時間或資源需求)Where,when&in what quantities entities get routed nex
29、tWhere,when&in what quantities entities get routed next (實體於何地，何時並以何種數量前進至下站實體於何地，何時並以何種數量前進至下站)Time&resource requirements for moving to the next location Time&resource requirements for moving to the next location (移動至下一站的活動或作業的時間或資源需求移動至下一站的活動或作業的時間或資源需求)15Data Collection and AnalysisEntity Flow Di
30、agram for Patient Processing(病患處理過程之實體流程圖)16Data Collection and AnalysisProcess Description for Patient Processing(病患處理過程之過程敘述表)17Data Collection and AnalysisDefining Incidental Details(定義附帶的細節)n nIncidental data(downtimes,setups&work Incidental data(downtimes,setups&work priority)are not essential
31、but necessary in order priority)are not essential but necessary in order to have a complete&accurate model to have a complete&accurate model (附帶的細節非必要，如故障時間、裝置時間、工附帶的細節非必要，如故障時間、裝置時間、工作優先順序等，但昰若要完成一個正確模式是有作優先順序等，但昰若要完成一個正確模式是有必要的必要的)n nOnce a basic model constructed,any numerical Once a basic model
32、constructed,any numerical values(e.g.activity time,arrival rates.)should values(e.g.activity time,arrival rates.)should be firmed up be firmed up (一旦基本模型已建立，任何數值資料如活動時間、一旦基本模型已建立，任何數值資料如活動時間、到達率等應被強化到達率等應被強化 )18Data Collection and AnalysisMaking Assumptions(建立假設)n nSimulation cant run with incomplet
33、e data,so assumptions are required for any unknown future conditions(模擬無法執行不完全資料，對於未確定狀況應建立假設)n nAssumption must make sense in the overall operation of the model.Seeing absurd behavior may tell us that certain assumptions dont make sense(建立假設應合建立假設應合理，異常行為的發生可能是假設不合理理，異常行為的發生可能是假設不合理)19Data Collecti
34、on and AnalysisMaking Assumptions(建立假設)n nSimulation cant run with incomplete data,soSimulation cant run with incomplete data,so assumptions are required for any unknown future assumptions are required for any unknown future conditions conditions (模擬無法執行不完全資料，對於未確定狀況應建立假設模擬無法執行不完全資料，對於未確定狀況應建立假設)n n
35、Assumption must make sense in the overall Assumption must make sense in the overall operation of the model.Seeing absurd behavior operation of the model.Seeing absurd behavior may tell us that certain assumptions dont make may tell us that certain assumptions dont make sense sense (建立假設應合理，異常行為發生可能是
36、假設不合理建立假設應合理，異常行為發生可能是假設不合理)20Data Collection and AnalysisMaking Assumptions(建立假設)n nSensitivity analysis assess the influence of an assumption on the validity of a model.(敏感性分析可以用來評估假設對模型的影響敏感性分析可以用來評估假設對模型的影響)Best or most optimistic case Best or most optimistic case(最樂觀情形最樂觀情形)Worst or most pessim
37、istic case(Worst or most pessimistic case(最悲觀情形最悲觀情形)Most likely or best guess case(Most likely or best guess case(最可能情形最可能情形)21Data Collection and AnalysisStatistical Analysis of Numerical Data(數值資料統計分析)n nData should be analyzed to ascertain their suitability for use.(資料應分析使用適合度)n nData characteri
38、stics:(資料特徵)Independence(randomness)Independence(randomness)(獨立性或隨機性獨立性或隨機性)Homogeneity(data from the same distribution)Homogeneity(data from the same distribution)(齊一性：是否來自相同分配齊一性：是否來自相同分配)Stationary(distribution of data no change over Stationary(distribution of data no change over time)time)(穩定性穩定
39、性 ：資料分配是否隨時間改變：資料分配是否隨時間改變)22Data Collection and AnalysisStatistical Analysis of Numerical Data(數值資料統計分析)n nStat:Fit in Promodel can automatically analyze&test data in a simulation (Stat:FitStat:Fit可以自動分析與測試模擬中的資料可以自動分析與測試模擬中的資料)n nParameters(常見統計參數常見統計參數)n nMean Mean(平均數平均數)the average of the datat
40、he average of the datan nMedian Median(中位數中位數)the value of middle the value of middle observationobservationn nMode Mode(眾數眾數)the value with greatest frequencythe value with greatest frequency23Data Collection and AnalysisDescriptive Statistics(敘述統計)n nParameters(常見統計參數常見統計參數)n nStandard Deviation S
41、tandard Deviation(標準差標準差)measure of measure of average deviationaverage deviationn nVariance Variance(變異數變異數)the square of standard the square of standard deviationdeviationn nCoefficient of variation Coefficient of variation(變異係數變異係數)standard standard deviation divided by mean deviation divided by
42、mean n nSkewness Skewness(偏態偏態)measure of symmetrymeasure of symmetryn nKurtosis Kurtosis(峰態峰態)measure of flatness or measure of flatness or peakednesspeakedness24Data Collection and AnalysisSuitability of Data for Use(資料適合度)n nTest for Independency(獨立性檢定):Data Data are independent if the value of o
43、ne observation is not are independent if the value of one observation is not influenced by the value of another observationinfluenced by the value of another observation (資料為獨立若其一觀察值不受其他觀察值的影響資料為獨立若其一觀察值不受其他觀察值的影響)n nTest for Homogeneity(齊一性檢定)：data data from the same distributionfrom the same distr
44、ibution (資料來自相同分配資料來自相同分配)n nTest for Stationary Data(穩定性檢定)：distribution of data does not change over timedistribution of data does not change over time (資料分配不隨時間改變而改變資料分配不隨時間改變而改變)25Data Collection and AnalysisTest for Independency(檢定資料獨立性的方法)Scatter Plot(分散點圖)A plot of adjacent points in the sequ
45、ence of A plot of adjacent points in the sequence of observed values plotted against each otherobserved values plotted against each otherA pair of consecutive observations(Xi,Xi+1),A pair of consecutive observations(Xi,Xi+1),i=1,.,n-1 i=1,.,n-1(一連串連續觀察值一連串連續觀察值)XisXis Positively correlated(Positivel
46、y correlated(正相關正相關)positively sloped trend line positively sloped trend line(正斜率直線正斜率直線)XisXis Negatively correlated(Negatively correlated(負相關負相關)Negatively sloped trend line Negatively sloped trend line(附斜率直線附斜率直線)26Data Collection and AnalysisTest for Independency(檢定資料獨立性的方法)Autocorrelation Plot(
47、自相關性)If observations in a sample are independent,they are uncorrelated.(若觀察值獨立則不相關)Assume that data are taken from stationary processThe measure of autocorrelation is called rho()(see,p.104)(自相關測量值稱為)Autocorrelation is between-1,1.(-1=1)If is near either extreme 1 or-1,the data is auto-correlated.(越
48、靠近1 or-1，則自相關越強)If is near 0,the data is little or unrelated (越靠近0，則相關性越弱)27Data Collection and AnalysisTest for Independency(檢定資料獨立性的方法)Runs Test(執行測試)A run in a series of observations is the occurrence of an uninterrupted sequence of numbers showing the same trend e.g run“up”or“down”;(顯現相同趨勢之序列，如向
49、上或向下走勢)28Data Collection and AnalysisTest for Independency(檢定資料獨立性的方法)Types of runs tests:if there are too many or too few,the randomness of the series is rejected.(趨勢出現次數過多，則應棄卻隨機性假設)nMedian Test(中位數檢定法):measure the number of runs(sequences of numbers)above and below the mediannTurning Point Test(轉
50、折點檢定法):measure the number of times the series changes directions29Data Collection and AnalysisTest for Homogeneity(齊一性檢定)n nTest for Identically Distributed Data):Test for Identically Distributed Data):Test if data Test if data set come from the same distribution.set come from the same distribution.