1.假设检验介绍--6sigma.pptVIP

正确决策 II类错误 I类错误 正确决策 零假设真 零假设伪 不否定零假设 否定零假设 真实情况 I 类和II类错误 决策 在法庭上，首先假设被告无罪，直到证明他有罪。假设检验类似于： H0: 被告无罪 H1: 被告有罪 错误举例 我们的法律系统也曾经作出错误的判决。理想的情况是，真正有罪的时候，我们就宣布有罪，无罪的时候就宣布无罪。 以下情况是什么类型错误 当被告无罪的时候，陪审团得出有罪结论？ 当被告有罪的时候，陪审团得出无罪结论？ 零假设真 零假设伪 不否定零假设 拒绝原假设 决策 正确决策 I类错误 II类错误 正确决策 H0: 被告无罪 H1: 被告有罪 真实情况 I类错误和II类错误 当零假设真时，否定零假设的错误。 I类错误发生的概率叫做显著水平，由?代表。 常见设定: ? = 0.05 I类错误发生的概率 检出能力是正确否定错误的零假设的概率 检出能力是1减去β错误 检岀能力 置信度 检出能力 检验总体均值是否等于目标值 检验两个总体均值是否相等 检验两个以上总体均值是否相等 检验方差是否相等 检验两个总体比率是否相等 检验关联性(多比例) 假设检验的几种情况 1 样本 Z 检验 1 样本 T 检验 2 样本 T 检验 配对 T 检验 ANOVA 相关性分析 回归分析 Chi-方 检验 1 比率检验 2 比率检验 对数回归(逻辑回归) Y连续 X连续 Y离散 X连续 Y连续 X离散 Y离散 X离散 D M A I C 常用的推理统计分析工具表 Hypothesis Testing Johnson Controls, Inc. ? 9-19-2000 W2HypothesisTesting.ppt * Hypothesis Testing Johnson Controls, Inc. ? 9-19-2000 W2HypothesisTesting.ppt * This graphical presentation of the data suggests that there is a difference between the two lines’ output performance. Six Sigma principles require, however, that quantitative evidence be evaluated by quantitative methods (statistics!) in order to make that decision. The difference in the lines’ outputs for this example might be intuitively obvious to you, but what if the difference is not so (visibly) apparent? Hypothesis Testing Johnson Controls, Inc. ? 9-19-2000 W2HypothesisTesting.ppt * This illustrates the difficulty in using visual, intuitive weighting to make decisions. What if there were more observations, what if there were fewer, how large does the difference really need to be, etc. There are two mistakes that can be made in answering the question as to whether the lines being compared are really different: You decide that there is a difference between the two lines, but there really is no difference. You decide that there is no difference between the two lines, but there really is a difference. In general, the same two mistakes can be made in any comparative decision of this sort (ii.e., a binomial equal/not-equal, greater/not-greater, yes/