Figure 5.6 Distribution consisting of two clusters.
Interrelationships between parameters must be eliminated before a correlation can be considered acceptable, for two reasons. Firstly, the use of highly correlated parameters makes mechanistic interpretation of the QSAR difficult; secondly high intercorrelation can render the statistical analysis unstable. Before an expression of the form of Equation [5.42] can be considered valid, linear regressions should be carried out between each pair of parameters. This is best expressed in the form of a correlation matrix, an example of which is given in Table 5.11.
Was this article helpful?