minitab数据处理分析培训教程

MINITAB上課教材永順by2008.10.11updata目錄1.Graph•直方圖•柏拉圖•管制圖•Cpk檢定2.檢定•2-samplet檢定(常態檢定,變異數檢定)•Anova檢定•2-Proportions3.MSA•GateR&R(計量)•GateR&R(計數)4.DOE•FactorialDOE5.範例操作介面指令區結果區數據區欄位名稱1.1Histogram(直方圖)選擇GraphHistogram選擇項目可以多個1.2ParetoChart(柏拉圖)選擇StatQualitytoolsParetoChartLabelsin:設定為defectmodeFrequenciesis:設定為ppmdata1.3Xbar-R(管制圖)選擇StatControlChartsVariablesChartsforSubgroupsXbar-R每群組samplesize1.4ProcessCapability(制程能力分析)選擇StatQualityToolsCapabilityAnalysisNormal規格下限規格上限2.1.12-sampleTTest(NormalityTest)選擇StatBasicStatisticsNormalityTestF1P-value＜0.05,為非常態分佈,F2之P-value＞0.05,為常態分佈先檢定F1之常態,再選擇另外一項目2.1.22-sampleTTest(VariancesTest)選擇StatBasicStatistics2Variances變異數分佈比對•若二者都是常態則看F-Test判定,否則看Levene’sTest判定:•F1P-value＜0.05,變異數有差異,F2之變異數檢定無差異2.1.32-sampleTTest選擇StatBasicStatistics2-SampleTestP-value0.05,二者無差異,表示改善前後比對沒有差異變異數相等則打勾2.2.1ANOVATest選擇StatANOVAOne-way•單因子多變數用ANOVAone-way分析•P-value＞0.05,ANOVA檢定並無顯著差異2.2.2ANOVATest選擇StatANOVATwo-way•兩因子多變數用ANOVATwo-way分析•Temp.anditemP-value均＞0.05,表明這兩個因子並非顯著的因子•Interaction交互作用比對,P-value0.05為有交互作用的影響2.32-Proportions選擇StatBasicStatistics2-ProportionsP-value＜0.05,二者有差異,表示此兩組數據有差異Trials:samplesize,Events:Ngpcs2.4.1ANOVAtwo-way範例•取同批產品FQC留樣及Tapping後產品，分別于酒精清洗前后做WettingBalance實驗，實驗數據如下表•用ANOVA方法分析留樣品與Tapping後產品清洗前後是否有差異T2/3F1F2T2/3F1F2T2/3F1F2T2/3F1F20.640.420.560.160.810.830.160.680.680.150.750.820.210.750.740.210.800.770.150.740.720.160.800.800.150.830.830.170.790.790.160.740.720.160.700.690.160.900.860.260.800.780.150.730.690.140.690.700.380.810.840.210.830.770.140.700.690.150.710.720.220.710.710.250.960.920.170.780.720.140.650.650.160.760.760.320.880.900.180.780.760.160.710.680.150.730.720.190.900.890.180.730.720.170.730.740.200.690.620.160.860.900.150.740.700.160.720.750.140.760.770.201.020.960.190.810.820.180.860.85FQC清洗後FQC清洗前Tapping清洗前Tapping清洗後2.4ANOVAtwo-way範例取樣P0.05，為顯著因子﹔處理P＞0.05，非顯著因子﹔InteractionP＞0.05,沒有交互影響作用F1、F2判定方法同T2/33.1MSA(計量分析)選擇StatQualityToolGageStudyGageR&RStudy(Crossed)TotalGateR&R之判定1.10%為可接受之量測系統2.10%~30%為接受度需視量測系統情況而定3.30為不能接受之量測系統3.2MSA(計數分析)選擇StatQualityToolAttributeAgreementAnalysisTotalGateR&R之判定方式1.=90%為可接受之量測系統2.=80%為接受度,但須改進3.80%為不能接受之量測系統4.1DOE選擇StatDOEFactorialCreateFactorialDesign按直交表因子填上實驗結果重複次數因子數4.2DOE選擇StatDOEFactorialAnalyzeFactorialDesign超過此線為顯著因子紅色點為顯著因子4.3DOE選擇StatDOEFactorialFactorialPlots圖形結果見下頁說明主效應圖交互作用圖4.4DOE選擇StatDOEFactorialFactorialPlots主效應圖為望小特性,故17A,1.9um,790度為較佳的條件,兩條線平行無交互作用,不平行直線則可能有交互作用4.4.1DOE二因子範例•三種不同材料在三種不同使用環境下量測電池壽命，每一種處理各重復四次，共實驗36次,實驗數據請見下表材料種類(M)溫度(T)15sum70sum125sumyb.A130155539344022920702309987418080758258B150188623136122479257019813001591261061155845C1381105761741205819610434215011681601501398260ya.173812917703799實驗結果紀錄入Result4.4.2DOE二因子範例選擇3Levels重覆四次選擇ANOVATableANOVA分析結果溫度與材料兩因子,P值均0.05為顯著的因子,且有交互作用的影響4.4.3DOE二因子範例根據望大特性,溫度15℃,第3種材料是最佳因子條件3條線彼此相交,說明有交互影響作用4.4.4DOE二因子範例1.ImprovedNPOCapdrop2.Curing升溫10度前後WB結果比對範例測試題目一•請用ANOVA與DOE分析改善前後與溫度之間的差異item215度225度235度改善前0.750.770.86改善前0.680.770.82改善前0.710.80.84改善前0.710.820.79改善前0.580.760.86改善前0.720.790.85改善前0.740.770.92改善前0.680.820.82改善前0.70.770.82改善前0.70.840.78改善後0.80.820.78改善後0.860.790.91改善後0.90.840.71改善後0.930.760.83改善後0.910.860.78改善後0.910.860.8改善後0.880.80.79改善後0.930.840.82改善後0.960.80.85改善後0.90.810.73•化學反應中濃縮的百分比，反應槽的壓力及反應時間等三個因子是被認為影響產品強度的最主要因子。選出濃縮3個水准，壓力3個水准及時間3個水准，進行重復兩次的因子實驗并得到以下資料：•請用DOE分析下表結果濃縮百分比時間3小時時間4小時壓力壓力4005006004005006002196.6197.7199.8198.4199.6200.6196196199.8198.6200.4200.94198.5196198.4197.5198.7199.6197.2196.9197.6198.11981998197.5195.6197.4197.6197198.5196.6196.2198.1198.4197.8199.8測試題目二