Classification by the SIMCA method

It is often the case that several variables are used to characterize a set of objects and that these can be divided into natural and disjunct subgroups. It can reasonably be assumed that members of such subgroups are similar to each other, but are more or less different to members of other subgroups. Such subgroups is hereafter called classes. Examples of such classes are Analytical samples of different origins different batches of raw material different substitution patterns in chemical compounds, e.g. cis/trans isomers clinical data of healthy and ill persons. 

We may wish to use an a priori known classification of the objects to determine to which class a new object belongs. We may also wish to know in which respect the classes are similar to each other and in which respect they are different. 

It is seen that a number of common problems can be regarded as classification problems. 

In the SIMCA method for classification,[15,16] separate principal components models are determined for each class. The idea behind this is that classes are fairly homogeneous and that the objects in a class are similar to each other, and it is very likely that a principal components model with few components is sufficient to describe the variation within a class. When a new object is projected down to the 

It is not within the scope of this book to go into any details on the SIMCA technique. Readers who wish to have more of technical details should consult the literature given in the reference list.[15—17] 

---

Vogt NB, Brakstad F, Thrane K, et al. 1987. Polycyclic aromatic hydrocarbons in soil and air Statistical analysis and classification by the SIMCA method. Environ Sci Technol 21 35-44. 

---