You are here:
COMPARISON OF SIMCA PATTERN RECOGNITION & LIBRARY SEARCH IDENTIFICATION OF HAZARDOUS COMPOUNDS FROM MASS SPECTRA
Citation:
Sarker, M., W. Glen, L. Yin, W. Dunn, III, D. Scott, AND S. Swanson. COMPARISON OF SIMCA PATTERN RECOGNITION & LIBRARY SEARCH IDENTIFICATION OF HAZARDOUS COMPOUNDS FROM MASS SPECTRA. U.S. Environmental Protection Agency, Washington, D.C., EPA/600/J-92/294.
Description:
SIMCA pattern recognition methods have been applied to mass spectral data from a target list of hazardous chemicals. cheme has been proposed for classification and identification of five classes of compounds including aromatics, chlorocarbons, bromocarbons, hydrocarbons, and polychlorinated biphenyls, PCBs. n addition, partial least squares regression has been used to predict the number of chlorine atoms present in the PCBs. raining set of 429 compounds was used. lassification models were derived by using principal components analysis of the autocorrelation transformed spectra and were applied to two GC/MS ambient air field samples and one hazardous waste sample. n optimal number of three principal components was determined. or identification the mass spectra of the unknowns were compared with the training set mass spectra in the predicted class. he overall classification and identification rates for training spectra were 89% and for the unknowns were 92% and 82%, respectively. dentification results of the SIMCA scheme for an ambient air field sample also was compared to that of PBM. orrect identification for the spectra in this sample by SIMCA was 71/84 (85%) as opposed to 35/84 (42% by PBM.