Intel 比较分析.ppt

* * * * * * * * * * * * * * * * * * * * * * Comparative Analysis Rev 1 Jul 04 * Comparative Analysis Rev 1 Jul 04 Comparative Analysis (WP) Comparative Analysis Introduction Type of Hypothesis Testing WP - Experiment Design WP - Sample Size WP – Analysis Method WP – Analysis Result Important Point Good Data Analysis ? Engineering Knowledge + Statistical Knowledge Only Engineering Knowledge Only Statistical Knowledge Based on My data analysis, if we reduce the number of apartment, we will reduce the accident on the road. The machine does not operate this morning, it must be the root cause for the problem encountered last weekend. Statistical Hypothesis Testing Goal : Collect data from an experiment/evaluation and make a decision based on the data Decision : Reject the null hypothesis (NOT SE) / Fail to reject the null hypothesis (SE) JMP : Prob .... 0.05 ? Do not reject H0 ; Statistically Equivalence (SE) Prob .... 0.05 ? Reject H0 ; Not SE (SW / SB) Terminology SE : Statistically Equivalence The prob .... 0.05 Not SE : Not Statistically Equivalence The Prob ... 0.05 SB : Statistically Better. NEW is better than POR. SW : Statistically Worse. NEW is worse than POR TE : Technically Equivalence. NEW is worse than POR statistically but the difference is within acceptable level. Need to have the change team agreement on TE. POR – Plan of Record Possible Outcomes Consequences Type 1 error (a) : Incorrectly reject H0 (when H0 is true) Type 2 error (b) : Falling to reject H0 (when H0 is not true) Example of the risk a Risk (Decision – Not SE ; Actual - SE) We concluded that there would be an improvement in the delivery lead time by switching to another forwarder company. However, after engaging the new forwarder for a month, improvements were not observed b Risk (Decision - SE ; Actual – Not SE) We concluded that there is no difference in the quality of goods delivered by both vendor A and B based on the evaluation done. However, after ordering from bo