||Numerical Index for Characterizing Data Set Separation.
Lafemina, D. H. ;
Jurs, P. C. ;
||Pennsylvania State Univ., University Park.;Health Effects Research Lab., Research Triangle Park, NC.
Pattern recognition ;
Organic compounds ;
Numerical analysis ;
Molecular structure ;
Chemical reactivity ;
Data sets ;
||Most EPA libraries have a fiche copy filed under the call number shown. Check with individual libraries about paper copy.
A method is reported for assessing the degree of separation between two clusters of points representing chemical data. The method, based on trend vectors, has been tested on both randomly generated and actual data sets. The index is intended to be a quick method for detecting the relative degree of separation between data sets (i.e., suitability of descriptors being used) in structure-activity studies or other pattern-recognition studies.