You are here:
APPLICATION OF SIMCA (SOFT INDEPENDENT MODELING OF CLASS ANALOGY) PATTERN RECOGNITION TO AIR POLLUTANT ANALYTICAL DATA
Citation:
Scott, D. APPLICATION OF SIMCA (SOFT INDEPENDENT MODELING OF CLASS ANALOGY) PATTERN RECOGNITION TO AIR POLLUTANT ANALYTICAL DATA. U.S. Environmental Protection Agency, Washington, D.C., EPA/600/D-84/271 (NTIS PB85124683).
Description:
The SIMCA 3B computer program is a modular, graphics oriented pattern recognition package which can be run on a microcomputer with limited memory, e.g. an Osborne 1 with 64K memory. Principal component analysis is used to classify data with this program. The SIMCA program was used to display small analytical data sets for exploratory data analysis after principal component fitting. K-Nearest Neighbor distances were also computed. The data sets included an interlaboratory comparison of trace element analyses of particulates by X-ray emission; a comparison of two methods for analyzing lead in gasoline; and GC/MS analysis of volatile organic compounds in ambient air. A combination of principal component and K-Nearest Neighbor analysis was found to provide a convenient and quick method for detecting outliers in the data sets.