Minitab常用操作说明VIP免费

ConfidentialMinitab常用操作說明Revision:APreparedby:蔡培奪Date:12/1‘2006Confidential內容：1.產品尺寸評估方式說明2.客訴報告中常用操作(2-samplet）3.GaugeR&R評估說明Confidential1.以Boxplot確認量測值內是否有離散值2.以NormalityTest確認量測值是否呈常態分配3.以1-sampleT檢定量測值是否與目標值相符4.以CapabilityAnalysis來確認Cpk&SigmaLevel1.產品尺寸評估方式說明Confidential1.1.Boxplot離散值確認Measure3.1623.1603.1583.1563.1543.1523.150BoxplotofMeasuremeasure4.2304.2254.2204.2154.2104.205Boxplotofmeasure有離散值無離散值•GraghBoxplotOneY/Simple6227產品尺寸評估報告Confidential1.1.Boxplot離散值確認從Boxplot確認是否有Outlier,若有的話需再確認其是否已outspec(有無異常).Confidential1.2.量測值是否呈常態分配確認•StatBasicStatisticNormalitytestConfidentialMeasurePercent3.1643.1623.1603.1583.1563.1543.1523.150999590807060504030201051Mean0.1253.156StDev0.002893N32AD0.574P-ValueProbabilityPlotofMeasureNormalmeasurePercent4.2304.2254.2204.2154.2104.2054.200999590807060504030201051Mean<0.0054.217StDev0.005622N32AD1.572P-ValueProbabilityPlotofmeasureNormalP-value>0.05,量測值呈“常態分配”P-value<0.05,量測值非“常態分配”通過P-value是否大於0.05來判定是否為常態分布.Confidential1.3.以1-sampleT檢定量測值是否與目標值相符此部分填入目標值(Nominalvalue)•StatBasicStatistics1-SampletConfidential確認P-value是否>0.05,若其>0.05即表示其接受虛無假設Ho,也就是量測值等於目標值;若P-value<0.05即表示拒絕虛無假設而接受對立假設,也就是量測值不等於目標值.Confidential1.4.以CapabilityAnalysis來確認Cpk&SigmaLevel首先需先確認其是否為常態分配(同1.2量測值是否呈常態分配確認）MeasurePercent3.1643.1623.1603.1583.1563.1543.1523.150999590807060504030201051Mean0.1253.156StDev0.002893N32AD0.574P-ValueProbabilityPlotofMeasureNormalmeasurePercent4.2304.2254.2204.2154.2104.2054.200999590807060504030201051Mean<0.0054.217StDev0.005622N32AD1.572P-ValueProbabilityPlotofmeasureNormalConfidential1.4.1.當量測值為常態分配•StatQualityToolsCapabilityAnalysisNormalConfidential由於我們為一個lot收集32個量測值而非不同lot進行收集,因此於Subgroupsize部分填入32.分別填入既定的尺寸上下限值ConfidentialConfidential0.7200.7080.6960.6840.6720.6600.648LSLUSLProcessDataSampleN32StDev(Within)0.00895598StDev(Overall)0.00895598LSL0.64Target*USL0.72SampleMean0.683906Potential(Within)CapabilityOverallCapabilityZ.Bench4.03Z.LSL4.90Z.USL4.03Ppk1.34CpmZ.Bench*4.03Z.LSL4.90Z.USL4.03Cpk1.34ObservedPerformancePPMUSL0.00PPMTotal0.00Exp.WithinPerformancePPMUSL27.87PPMTotal28.35Exp.OverallPerformancePPMUSL27.87PPMTotal28.35WithinOverallProcessCapabilityofC1可分別得到Cpk&ZBench•若數據為短期：CPK=1.34sigmalever=Z.Bench=4.93•若數據為長期：CPK=1.34sigmalever=Z.Bench+1.5=6.43ConfidentialConfidential1.4.2.當量測值為非常態分配填入組內量測值數量•StatControlChartsBox-CoxTransformticConfidentialLambdaStDev5.02.50.0-2.5-5.00.0059750.0059500.0059250.0059000.0058750.005850Lambda-5.00(using95.0%confidence)Estimate-5.00LowerCL*UpperCL*RoundedValueBox-CoxPlotofC1正態轉換係數Lambda=-5.00Confidential將之前得到的正態轉換係數填入即可,其他步驟則與量測值為常態分配時一樣•StatQualityToolsCapabilityAnalysisNormalBoxCoxConfidential0.04680.04560.04440.04320.04200.0408LSL*USL*transformeddataProcessDataSampleN32StDev(Within)0.00591975StDev(Overall)0.00591975AfterTransformationLSL*0.0474145Target*LSL*USL*0.0403861SampleMean*0.0430389...