minitab数据

woshixingyan

请各位大侠帮我看看这个模型行不行，其中S=20.89和S=2.778是什么意思呐？

试验设计：三因素五水平的试验
            
                               试验所得Ta的值
-2    0    0    223.77
0    0    2    329.96
1    -1    1    243.71
0    0    -2    114.60
-1    -1    -1    192.15
1    -1    -1    346.43
1    1    -1    192.40
0    0    0    459.81
1    1    1    469.98
2    0    0    375.27
-1    1    -1    166.08
-1    1    1    456.16
-1    -1    1    154.93
0    -2    0    97.79
0    2    0    315.30


The analysis was done using coded units.

Estimated Regression Coefficients for Ta

Term         Coef  SE Coef       T      P
Constant  442.647   18.422  24.028  0.000
A          36.637    5.222   7.016  0.001
B          48.901    5.222   9.364  0.000
C          53.653    5.222  10.274  0.000
A*A       -37.927    6.276  -6.043  0.002
B*B       -61.171    6.276  -9.746  0.000
C*C       -57.237    6.276  -9.120  0.000
A*B       -25.365    7.385  -3.435  0.019
A*C        -9.750    7.385  -1.320  0.244
B*C        88.450    7.385  11.977  0.000

S = 20.89   R-Sq = 99.0%   R-Sq(adj) = 97.3%


Analysis of Variance for Ta

Source          DF  Seq SS  Adj SS   Adj MS      F      P
Regression       9  223301  223301  24811.2  56.86  0.000
  Linear         3  105796  105796  35265.2  80.82  0.000
  Square         3   49010   49010  16336.8  37.44  0.001
  Interaction    3   68495   68495  22831.6  52.33  0.000
Residual Error   5    2182    2182    436.3
Total           14  225482




Estimated Regression Coefficients for Ta using data in uncoded units

Term          Coef
Constant   442.647
A          36.6375
B          48.9013
C          53.6525
A*A       -37.9271
B*B       -61.1708
C*C       -57.2371
A*B       -25.3650
A*C       -9.75000
B*C        88.4500



Response Surface Regression: Ce versus A, B, C

The analysis was done using coded units.

Estimated Regression Coefficients for Ce

Term         Coef  SE Coef       T      P
Constant  10.7856   2.4497   4.403  0.007
A          0.0219   0.6944   0.032  0.976
B          0.2806   0.6944   0.404  0.703
C         -5.0044   0.6944  -7.207  0.001
A*A        3.2243   0.8346   3.863  0.012
B*B        2.3456   0.8346   2.810  0.038
C*C        1.9656   0.8346   2.355  0.065
A*B       -2.4062   0.9821  -2.450  0.058
A*C        1.4137   0.9821   1.440  0.210
B*C        0.7987   0.9821   0.813  0.453

S = 2.778   R-Sq = 93.8%   R-Sq(adj) = 82.7%


Analysis of Variance for Ce

Source          DF   Seq SS   Adj SS   Adj MS      F      P
Regression       9  586.912  586.912   65.212   8.45  0.015
  Linear         3  401.968  401.968  133.989  17.37  0.004
  Square         3  117.530  117.530   39.177   5.08  0.056
  Interaction    3   67.414   67.414   22.471   2.91  0.140
Residual Error   5   38.578   38.578    7.716
Total           14  625.489




Estimated Regression Coefficients for Ce using data in uncoded units

Term           Coef
Constant    10.7856
A         0.0218750
B          0.280625
C          -5.00438
A*A         3.22431
B*B         2.34556
C*C         1.96556
A*B        -2.40625
A*C         1.41375
B*C        0.798750

woshixingyan

A是由B降解而生成的，所以A的响应值无论如何都不会是负值，但是用这个模型在自变量选定值的两边却出现了负值，有人报道过这样的响应面图，中间有山峰形的鼓包，外围是和xy面平行的面，就像有帽檐的帽子一样，我觉得我的图也应该是这样的，但是自己不会做，哪位大侠能告诉我这样的图怎么做啊，模型需要怎样的修改，才能达到这样的效果，非常感谢！

细履平沙

S=20.89和S=2.778  第1个模型是针对A的。它的标准差是20.89
第2个模型是针对B的，它的标准差是2.778.
楼主将模型的响应设为A和B，而同时将因素也设为ABC，这样容易引起混淆。

0202010109

dddddddddddddd