24小时热门版块排行榜

返回列表

aagemini

铁虫 (小有名气)

应助: 0 (幼儿园)
金币: 789.4
红花: 9
帖子: 184
在线: 39小时
虫号: 2782202
注册: 2013-11-06
专业: 有机高分子功能材料

[求助] 用design expert优化制备工艺的问题求解求统计高手

第一次用design expert，本身数学也不好，摸摸爬爬写了篇文章，是优化制备条件的，选择最佳温度，时间，配比。
给老师看，老师的意见是：“最后数据统计分析还需要补充分析数据，根据模型最佳工艺条件应该推算出来”。
下图是他给的表，让我补充红色字体。

我一开始的模型里没有选择红色的这些项，因为我试了好多选择，只有这种情况时，p-value，Prob > F显著，而失拟值Lack of Fit不显著

当我按照老师的要求修改时，

方差分析如下：
Sum of Mean F p-value
Source Squares df Square Value Prob > F
Model 89.81 12 7.48 50.39 0.0009 significant
  A-reaction?time 0.12 1 0.12 0.80 0.4223
  B-temperature 3.57 1 3.57 24.04 0.0080
  C-reactants ratio 2.00 1 2.00 13.49 0.0213
  AB 13.85 1 13.85 93.23 0.0006
  AC 0.48 1 0.48 3.21 0.1475
  BC 0.099 1 0.099 0.67 0.4603
  A^2 0.46 1 0.46 3.07 0.1545
  B^2 2.16 1 2.16 14.53 0.0189
  C^2 11.93 1 11.93 80.33 0.0009
  A^2B 10.79 1 10.79 72.66 0.0010
  A^2C 0.59 1 0.59 3.99 0.1163
  AB^2 2.56 1 2.56 17.23 0.0142
  AC^2 0.000 0
  B^2C 0.000 0
  BC^2 0.000 0
Pure Error 0.59 4 0.15
Cor Total 90.40 16

The Model F-value of 50.39 implies the model is significant.  There is only
a 0.09% chance that a "Model F-Value" this large could occur due to noise.

Values of "Prob > F" less than 0.0500 indicate model terms are significant.
In this case B, C, AB, B++2+-, C++2+-, A++2+-B, AB++2+- are significant model terms.
Values greater than 0.1000 indicate the model terms are not significant.
If there are many insignificant model terms (not counting those required to support hierarchy),
model reduction may improve your model.

Std. Dev. 0.39 R-Squared 0.9934
Mean 2.48 Adj R-Squared 0.9737
C.V. % 15.54 Pred R-Squared N/A
PRESS N/A Adeq Precision 30.433
Case(s) with leverage of 1.0000:  Pred R-Squared and PRESS statistic not defined

"Adeq Precision" measures the signal to noise ratio.  A ratio greater than 4 is desirable.  Your
ratio of 30.433 indicates an adequate signal.  This model can be used to navigate the design space.

Coefficient Standard 95% CI 95% CI
Factor Estimate df Error Low High VIF
Intercept 2.78 1 0.17 2.30 3.26
A-reaction?time -0.17 1 0.19 -0.71 0.36 2.00
B-temperature 0.94 1 0.19 0.41 1.48 2.00
C-reactants ratio 0.71 1 0.19 0.17 1.24 2.00
AB -1.86 1 0.19 -2.40 -1.33 1.00
AC -0.35 1 0.19 -0.88 0.19 1.00
BC -0.16 1 0.19 -0.69 0.38 1.00
A^2 0.33 1 0.19 -0.19 0.85 1.01
B^2 0.72 1 0.19 0.19 1.24 1.01
C^2 -1.68 1 0.19 -2.20 -1.16 1.01
A^2B 2.32 1 0.27 1.57 3.08 2.00
A^2C 0.54 1 0.27 -0.21 1.30 2.00
AB^2 -1.13 1 0.27 -1.89 -0.37 2.00
AC++2+-  ALIASED A, AB++2+-
B++2+-C  ALIASED C, A++2+-C
BC++2+-  ALIASED B, A++2+-B

=============================================================
                     W A R N I N G
   The model you selected has terms that are aliased with one another.
If you continue, the least squares parameter estimates for aliased models
will not be unique and the resulting contour plots will be misleading.
=============================================================

Final Equation in Terms of Coded Factors:

electrical conductivity    =
+2.78
-0.17    * A
+0.94    * B
+0.71    * C
-1.86    * A * B
-0.35    * A * C
-0.16    * B * C
+0.33    * A^2
+0.72    * B^2
-1.68    * C^2
+2.32    * A^2 * B
+0.54    * A^2 * C
-1.13    * A * B^2

Final Equation in Terms of Actual Factors:

Not available for ALIASED models.

The Diagnostics Case Statistics Report has been moved to the Diagnostics Node.
In the Diagnostics Node, Select Case Statistics from the View Menu.

Proceed to Diagnostic Plots (the next icon in progression).  Be sure to look at the:
1) Normal probability plot of the studentized residuals to check for normality of residuals.
2) Studentized residuals versus predicted values to check for constant error.
3) Externally Studentized Residuals to look for outliers, i.e., influential values.
4) Box-Cox plot for power transformations.

If all the model statistics and diagnostic plots are OK, finish up with the Model Graphs icon.

请问这样的模型可以用吗？

回复此楼