Re: how to analyze the orthogonal experiment data with strong dependence.
- From: Jinsong Zhao <jinsong.zhao@xxxxxxxxx>
- Date: Mon, 28 Sep 2009 21:12:10 -0700 (PDT)
On Sep 29, 12:03 am, "w.ccarleton" <w.ccarle...@xxxxxxxxx> wrote:
I'm not an expert, but if you're planning on modeling the data with
regression you can use Generalized Least Squares (either linear or non-
linear) in order to do so. The GLS approach is designed to be used in
cases wherein there is a correlation between errors or data values.
You can use functions in R (http://www.r-project.org/) to do this with
relative ease. See also:
Takeaki Kariya and Hiroshi Kurata [2004] Generalized least squares.
Wiley series in probability and statistics. John Wiley and Sons.
On Sep 29, 12:03 am, "w.ccarleton" <w.ccarle...@xxxxxxxxx> wrote:
I'm not an expert, but if you're planning on modeling the data with
regression you can use Generalized Least Squares (either linear or non-
linear) in order to do so. The GLS approach is designed to be used in
cases wherein there is a correlation between errors or data values.
You can use functions in R (http://www.r-project.org/) to do this with
relative ease. See also:
Takeaki Kariya and Hiroshi Kurata [2004] Generalized least squares.
Wiley series in probability and statistics. John Wiley and Sons.
Thank you very much for your reply. I read the help of gls() from nlme
package.
If I'm right, gls() is used for the data, which errors are allowed to
be correlated
and/or have unequal variances. However, in my experiment, the
dependent variable
(Y in the following data) are strong correlated with experimental
factors (B and C).
In order to make my problem more clear. The following real
experimental results are given.
The A, B, C, D are experimental factors. X is the observed results,and
heavily
dependent on C, increasing with C. Y is the final results obtained
from X, and B and C
using the formula:
Y = X / B / C
In here, the complex coefficients were not given.
I hope to get the optimal experimental condition for my following
study. However, I can not
find a clue from these experimental data.
If the orthogonal design was not the correct design for those
situation, which design was the best choice?
Thanks in advance!
Jinsong
--------->8 experimental results start here 8<-----------
RUN A B C D X Y
1 4 5 0.01 15 0.115 134.578
2 4 10 0.1 20 0.218 12.727
3 4 20 0.25 25 0.264 3.084
4 4 30 0.5 30 0.252 0.984
5 4 60 1.0 35 0.261 0.255
6 5 10 0.01 25 0.054 31.743
7 5 20 0.1 30 0.183 5.354
8 5 30 0.25 35 0.195 1.521
9 5 60 0.5 15 0.248 0.484
10 5 5 1.0 20 0.251 2.934
11 6 20 0.01 35 0.048 13.970
12 6 30 0.1 15 0.114 2.223
13 6 60 0.25 20 0.185 0.720
14 6 5 0.5 25 0.128 2.996
15 6 10 1.0 30 0.206 1.205
16 7 30 0.01 20 0.085 16.628
17 7 60 0.1 25 0.200 1.950
18 7 5 0.25 30 0.200 9.362
19 7 10 0.5 35 0.168 1.960
20 7 20 1.0 15 0.244 0.713
21 8 60 0.01 30 0.116 11.288
22 8 5 0.1 35 0.117 13.721
23 8 10 0.25 15 0.156 3.657
24 8 20 0.5 20 0.203 1.189
25 8 30 1.0 25 0.325 0.633
--------->8 experimental results end here 8<-----------
.
- References:
- how to analyze the orthogonal experiment data with strong dependence.
- From: Jinsong Zhao
- Re: how to analyze the orthogonal experiment data with strong dependence.
- From: w.ccarleton
- how to analyze the orthogonal experiment data with strong dependence.
- Prev by Date: Re: Power analysis for 2-way ANOVA
- Next by Date: Re: standardize ordinal distribution.
- Previous by thread: Re: how to analyze the orthogonal experiment data with strong dependence.
- Next by thread: Re: how to analyze the orthogonal experiment data with strong dependence.
- Index(es):
Relevant Pages
|
