【正文】
Interactions 交互作用 Notice how this effect pares to the main effects for A and B. Which effect is the largest?注意這個效果和 A、 B主效果的對比,哪個的效果最大? Time Temp A B AB Yield 70 () 145 () + 56 130 (+) 145 () 69 70 () 165 (+) 82 130 (+) 165 (+) + 58 Six Sigma Green Belt Training Design of Experiments (DoE) ? Analysis of Variance (ANOVA) 方差分析 ? Used to determine if a factor is statistically significant in changing the response. 用于判定因子對響應是否有統(tǒng)計上的顯著影響 ? Used to define the y=f(x) Equation 用于確定 y=f(x) 公式 Analytical解析 Factorial Fit: Y versus A, B 因子擬合: Y與 A、 B Estimated Effects and Coefficients for Y (coded units) Y( 以編碼為單位)的估計效果和系數(shù) Term要素 Effect效果 Coef系數(shù) Constant常數(shù) A B A*B Y=79 + 2*A + *B 9*A*B Six Sigma Green Belt Training Design of Experiments (DoE) So what do you think happened to Alex? 這樣 Alex做了什么? ? There are a number of statistical software packages which aid in the design and analysis of DOE’s, providing both analytical and graphical results: 許多軟件用于幫助 DOE的設計和分析,提供解析和圖示的輸出結果: Case Study – Yield Improvement Section 6 案例分析 – 改善產出率之第六節(jié) Factorial Fit: Y versus A, B 因子擬合: Y與 A、 B Estimated Effects and Coefficients for Y (coded units) Y( 以編碼為單位)的估計效果和系數(shù) Term Effect Coef Constant A B A*B 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8ABA B效 果 的 柏 拉 圖P a r e t o C h a r t o f t h e E f f e c t s( r e s p o n s e i s Y , A lp h a = . 1 0 )A : AB : B8 18 29 56 0BA7 03 01 8 51 6 5Y 的 立 體 圖 (平 均 值 )C u b e P l o t ( d a t a m e a n s ) f o r YA B30701651857 27 68 08 48 8YY 主 因 子 圖 (平 均 值 )M a i n E f f e c t s P l o t ( d a t a m e a n s ) f o r Y3 0 7 0 1 8 51 6 59 08 07 06 0BAMeanY 的 交 互 作 用 圖 (平 均 值 )I n t e r a c t i o n P l o t ( d a t a m e a n s ) f o r YSix Sigma Green Belt Training Design of Experiments (DoE) Summary 總結 ? Learning requires two necessary elements: a significant event and a perceptive observer. 學習的兩個必要因素:顯著的事件和有洞察力的觀察者 ? Normal observation of a process may or may not allow an observer to witness the significant event. The event may simply never happen, or happen when no one is around to observe 可以目擊到顯著的事件。這個事件可能永遠不會發(fā)生,或者發(fā)生時卻沒人在旁邊看到。 ? Learning is accelerated by creating significant events (kicking the process by altering a number of independent variables and observing the results) . Designed experiments are the most effective way of doing this. 制造顯著事件能使學習加速(改變因變量使過程變化并觀察結果)。實驗設計是這樣做最有效的方法。 ? OFAT and trialanderror methods are the least effective ways of experimenting, and rarely lead to significant process knowledge. 一次調整一個因子和反復實驗是最無效的實驗,幾乎不會獲取對過程的真正認識。 ? DOE is a rich topic – see your Black Belt for more 豐富,找你們的黑帶了解更多的信息。