STATISTICAL THINKING AND TECHNIQUE FOR QSAR AND RELATED STUDIES .1. GENERAL-THEORY

The two parts of this paper form a critique of a variety of statistica l techniques of actual or potential use in quantitative structure-acti vity relationship (QSAR) studies and related fields. Part I explores t he statistical thinking that is needed to underpin those techniques. E mphasis is placed on (a) the role of 'exchangeability' as an alternati ve to unrealistic statistical modelling and (b) the use of cross-valid ation to limit self-deception in the use of any particular technique. The problem of the almost unlimited range of molecular descriptors is seriously addressed. (Part II provides a concise critical review of me thods - some well-established and some new.)