Abstract
Discriminant analysis can be used to (1) find criteria to separate observations into groups and (2) to optimally assign a new observation to its correct group. It is rather exploratory in nature especially when causal relationships are not known. Twenty four selected organics were tested as corrosion inhibitors and both physical and structural features were encoded as criteria for discrimination. In addition, a narrower set of twelve nitrogen containing compounds was also studied. In both cases the carbon chain length was an important uncoded criterion. The physical factors such as boiling point, molecular weight, pKa, and molecular area were not as important as some of the structural parameters.