Multivariate methods
Univariate techniques, as the name implies, consider only one variable at a time. Multivariate, again as the name implies, means the use of more than one variable at once. Since variables usually contain information it should be obvious that the inclusion of more information in an analysis is likley to lead to a better understanding of the problem. This may not be so obvious, however, so a simple example may make the point as shown to the right (click on it to expand):
In this two variable plot a cluster of active compounds, marked A, are clearly separated from the surrounding Inactive compounds (I). Neither variable on its own would be able to separate these compounds but this view of the data shows a clear pattern. If the combination of two variables allows such an improvement imagine what happens when all 20, 50 or 500 variables are used at once.Typical multivariate methods are:
The properties and use of such multivariate techniques are well described in the statistical literature but are less well known in the chemical communtity. ChemQuest has extensive experience in the application of multivariate statistics to a range of chemical problems and can provide solutions ranging from advice to the complete treatment of problems.
