DoE复习笔记

Analysis procedure of one-way experiment

factor	n data	sum
$A_1$		$T_{A_1}$
$A_i$	$x_{ij}$	$T_{A_i}$
$A_a$		$T_{A_a}$
Total Sum		T

1. Calculate the correction term
   $$CT=\frac{(Sumofalldata{)}^2}{numberofalldata}$$
2. Compute the total sum of squares.
   $$\begin{gathered} S_T=\sum (Squaredofindividualdata)-CT \\ {f}_T=(numberofdata)-1 \end{gathered}$$
3. Calculate the variation of the factor.
   $$\begin{gathered} S_A=\sum \frac{(Sumof{A}_i^{\prime }sdata{)}^2}{(Numerof{A}_i^{\prime }sdata{)}^2}-CT \\ {f}_A=(numberoflevels)-1 \end{gathered}$$
4. Calculate the variation of the error
   $$\begin{gathered} S_e=S_T-S_A \\ {f}_e=f_T-f_A \end{gathered}$$
5. Create an analysis of variance table
   $aisthenumberoflevel,nisthenumberofreplication$

factor	$S$	$f$	$V$	$F_0$
$A$	$S_A$	$f_A=a-1$	$V_A=S_A/f_A$	$f_0=V_A/V_e(mark)$
$e$	$S_e$	$f_e=f_T-f_A=a(n-1)$	$V_e=S_e/f_e$	–
$T$	$S_T$	$f_T=an-1$	–	–

7. Perform the statistical estimation.
   • about the mark
     • The variance ratio is compared with the value of F table $F_{f_e}^{f_A}(\alpha )$
     • If the factor becomes significant at a significant level of $5\%(>F_{f_e}^{f_A}(0.05))$, mark * is added to the right of the variance ratio.
     • If the factor becomes significant at a significant level of $1\%(>F_{f_e}^{f_A}(0.01))$, add two marks * to the right of the variance ratio.
   • contribution ratio
     • $factorcontributionratio:{\rho }_{factor}=\frac{S_{factor}-f_{factior}\times V_e}{S_T}\times 100(\%)$
     • $errorcontributionratio:{\rho }_e=100-\sum \rho$
   • statistical estimation
     • estimation of mean $$(meanatthelevel)\pm t(f_e,0.05)\sqrt{\frac{V_e}{n}}$$
     • estimation of the difference between means $$\begin{gathered} (thedifferencebetweenmeans)\pm t(f_e,0.05)\sqrt{\frac{2V_e}{n}} \\ ({\bar{A}}_i-{\bar{A}}_j)\pm t(f_e,0.05)\sqrt{(\frac{1}{n_i}+\frac{1}{n_j})V_e} \end{gathered}$$
     • the existence range of individual data$$(meanatthelevel)$$

Two-way layout without replication

	$B_1.....B_j.....Bb$	sum
$A_1$		$T_{A_1}$
$A_i$	$x_{ij}$	$T_{A_i}$
$A_a$		$T_{A_a}$
Total Sum	$T_{B_1}.....T_{B_j}.....T_{B_b}$	T

• Calculate the correction term
  $$CT=\frac{(Sumofalldata{)}^2}{numberofalldata}$$
• Compute the total sum of squares.
  $$\begin{gathered} S_T=\sum (Squaredofindividualdata)-CT \\ {f}_T=(numberofdata)-1 \end{gathered}$$
• Calculate the variation of the factor A.
  $$\begin{gathered} S_A=\sum \frac{(Sumof{A}_i^{\prime }sdata{)}^2}{(Numerof{A}_i^{\prime }sdata)}-CT \\ {f}_A=(numberoflevels)-1 \end{gathered}$$
• Calculate the variation of the factor B.
  $$\begin{gathered} S_B=\sum \frac{(Sumof{B}_i^{\prime }sdata{)}^2}{(Numerof{B}_i^{\prime }sdata)}-CT \\ {f}_B=(numberoflevels)-1 \end{gathered}$$
• Calculate the variation of the error
  $$\begin{gathered} S_e=S_T-S_A \\ {f}_e=f_T-f_A \end{gathered}$$
• Create an analysis of variance table
  $aisthenumberoffactorA,bisthenumberoffactorB$

factor	$S$	$f$	$V$	$F_0$
$A$	$S_A$	$f_A=a-1$	$V_A=S_A/f_A$	$f_0=V_A/V_e(mark)$
$B$	$S_B$	$f_B=b-1$	$V_A=S_B/f_B$	$f_0=V_B/V_e(mark)$
$e$	$S_e$	$$\begin{gathered} f_e=f_T-f_A-f_B \\ =(a-1)(b-1) \end{gathered}$$	$V_e=S_e/f_e$	–
$T$	$S_T$	$f_T=ab-1$	–	–

8. Perform the statistical estimation

• Estimation of significant factor effects

• estimation of mean $$(meanatthelevel)\pm t(f_e,0.05)\sqrt{\frac{V_e}{numberofdataatthelevel}}$$

• estimation of the difference between means $$(thedifferencebetweenmeans)\pm t(f_e,0.05)\sqrt{\frac{2V_e}{numberofdataatthelevel}}$$

• Selection of optimal conditions and estimation of population mean
  1. If only one factor is significant, $T_{A_i}$is the optimal level, the population mean estimation is as follows.$$(meanateachlevel)\pm t(f_e,0.05)\sqrt{\frac{V_e}{numberofdataatthelevel}}$$
  2. If both factors are significant, is the optimal level combination, and the population mean estimate is obtained as follows$$\begin{gathered} \hat{\mu }(A_i{B}_j)\pm t(f_e,0.05)\sqrt{\frac{V_e}{n_e}}=\overline{A_i}+\overline{B_j}-\bar{T}\pm t(f_e,0.05)\sqrt{\frac{(a+b-1)V_e}{ab}} \\ {n}_e=\frac{ab}{a+b-1} \end{gathered}$$

• Estimate the difference of the mean

  • $Thedifferencebetweenthetwomeans\pm t(f_e,0.05)\sqrt{\frac{V_e}{n_d}}$
  • $n_d$ is called the effective number of replications of the mean difference
    $$\begin{gathered} whenthemeandifferencebetween{A}_i{B}_{j}and{A}_p{B}_q \\ (i\ne p,j\ne q)n_d=\frac{ab}{2(a+b)} \\ (i\ne p,j=q)n_d=\frac{b}{2} \\ (i=p,j\ne q)n_d=\frac{a}{2} \end{gathered}$$

• The existence range of individual data
  $(meanforeachlevelorcombinationoflevels)\pm t(f_e,0.05)\sqrt{(1+\frac{1}{n_e})V_e}$

Concept of estimation

Parameter	Estimation value
$\mu$	$\bar{T}$
$a_i$	${\bar{A}}_i-\bar{T}$
$b_i$	${\bar{b}}_i-\bar{T}$
$(ab{)}_{ij}$	$\overline{A_i{B}_j}-{\bar{B}}_j-{\bar{A}}_i+\bar{T}$

• In the case that only A is significant,$\hat{\mu }(A_i{B}_j)=\mu +a_i$ and $n_e=b$
• In the case that only B is significant,$\hat{\mu }(A_i{B}_j)=\mu +a_i$ and $n_e=b$
• In the case that both A and B are significant , $\hat{\mu }(A_i{B}_j)=\mu +a_i+b_j$ and $n_e=\frac{ab}{a+b}$

Two-way layout with replication

• Calculation of $T_{A_i{B}_j}$
   Create a table summing repetition data.
• Calculation of correction term（$CT$）
• Calculation of sum of squares
   $S_T,f_T$
• Calculation of the variation of the factor A
• Calculation of the variation of the factor B
• Calculation of the variation of the combination of A and B
  $$\begin{gathered} S_{AB}=\sum \sum \frac{(Sumof{A}_i{B}_j^{\prime }sdata{)}^2}{(Numberof{A}_i{B}_j^{\prime }sdata)}-CT \\ {f}_{ab}=ab-1 \end{gathered}$$
• Calculation of the variation of the interaction of A and B
  $$\begin{gathered} S_{A\times B}=S_{AB}-S_A-S_B \\ {f}_{A\times B}=f_{AB}-f_A-f_B=(a-1)(b-1) \end{gathered}$$
• Calculation of the variation of the error.
  $$\begin{gathered} S_e=S_{AB}-S_A-S_B-S_{A\times B}=S_T-S_{A\times B} \\ {f}_e=f_T-f_{AB} \end{gathered}$$
• Creation of an analysis of variance table

factor	$S$	$f$	$V$	$F_0$
$A$	$S_A$	$f_A=a-1$	$V_A=S_A/f_A$	$f_0=V_A/V_e(mark)$
$B$	$S_B$	$f_B=b-1$	$V_A=S_B/f_B$	$f_0=V_B/V_e(mark)$
$A\times B$	$S_{A\times B}$	$f_{A\times B}=(a-1)(b-1)$	$V_{A\times B}=S_{A\times B}/f_{A\times B}$	$f_0=V_{A\times B}/V_e(mark)$
$e$	$S_e$	$$\begin{gathered} f_e=f_T-f_A-f_B \\ =(a-1)(b-1) \end{gathered}$$	$V_e=S_e/f_e$	–
$T$	$S_T$	$f_T=ab-1$	–	–