Quartz Technology Ltd
|
Quartz Technology Ltd
|
|
|
Results of LDFA of the example data set are shown in Table 1.2. The success rate was found to be 98% using cross-validation (also known as leave-one-out) (Mertens et al. 1994).
Results of eigenanalysis for class and corresponding percent variance are shown in Table 1.3.
Table
1.3: Results of
eigenanalysis for class (MANOVA) using responses from sensors x1 - x5 .
The first two canonical variates account for a high percentage of variation in the data. The coefficients of the canonical variates are shown in Table 1.4. The values suggest the remaining sensors are not highly correlated.
Table
1.4 Coefficients of the
canonical variates.
Therefore, the equations generated through eigenanalysis for class (MANOVA) using responses from sensors x1 - x5 were: Z1 = -0.1628x1 - 0.1551 x2
- 0.2355x3 - 0.1145 x4 - 0.0372 x5 Z2 = 0.0986x1 + 0.1030 x2
+ 0.0815x3 + 0.1835 x4 + 0.0272 x5 A plot of Z1 versus Z2 is shown in Figure 1.8. The location of the propan-1-ol, ethanol and methanol cluster suggests similarities in vapour/coating interaction mechanisms. However, the propan-2-ol cluster was well separated from the main group of alcohols that may suggest other factors are also important, for example, the branched nature of this solute.
Figure 1 Application of MANOVA to responses from sensors x1 - x5 obtained at 50 seconds for (D) hexane, ( ÿ ) propan-2-ol, ( ÿ ) triethylamine, (O) acetone, ( ÿ ) chloroform, (´) diethylether, (D) toluene, (s) propan-1-ol, (D) 2-butanone, (O) ethanol, (´) acetonitrile, (´ ) methanol, (O) n--butylacetate. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Send mail to webmaster@quartztec.com with questions or comments about this
web site.
|
