Multivariate receptor modeling from a statistical science viewpoint /

Receptor modeling is a collection of methods used to model

Bibliographic Details
Main Author: Park, Eun Sug , 1967-
Format: Thesis Book
Language:English
Published: [Place of publication not identified] : [publisher not identified] ; 1997.
Subjects:
Online Access:http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=736824721&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD

MARC

Tag First Indicator Second Indicator Subfields
LEADER 00000ctm a22000005a 4500
001 in00001422321
005 20190325093517.0
008 980708s1997 xx a b 000 0 eng d
035 |a (OCoLC)39450058 
035 |9 AHR2007AM 
037 |a 98-15820  |b UMI 
040 |a TXA  |c TXA  |d UtOrBLW 
049 |a TXAM  |a TXAR 
099 |a 1997  |a Dissertation  |a P3752 
100 1 |a Park, Eun Sug ,  |d 1967- 
245 1 0 |a Multivariate receptor modeling from a statistical science viewpoint /  |c by Eun Sug Park. 
264 1 |a [Place of publication not identified] :  |b [publisher not identified] ;  |c 1997. 
300 |a xi, 153 leaves :  |b illustrations ;  |c 28 cm. 
336 |a text  |b txt  |2 rdacontent 
337 |a unmediated  |b n  |2 rdamedia 
338 |a volume  |b nc  |2 rdacarrier 
504 |a Includes bibliographical references: pages 143-147. 
500 |a Vita. 
502 |b Ph. D.  |c Texas A&M University  |d 1997. 
500 |a "Major Subject: Statistics". 
530 |a Issued also on microfiche from University Microfilms Inc. 
520 |a Receptor modeling is a collection of methods used to model  
520 |a the air pollution data. Although the main interest is  
520 |a identifying pollution sources, the determination of the  
520 |a number of major sources is the first problem that we  
520 |a should overcome. The NUMFACT algorithm is a new method of  
520 |a determining the number of underlying factors in a  
520 |a multivariate system. Although NUMFACT has not been  
520 |a previously described, it has been used in high profile  
520 |a studies. The asymptotic distribution of NUMFACT  
520 |a statistics and their associated cutoff points are derived.  
520 |a Modified NUMFACT statistics and related statistics, W, are  
520 |a also presented with a new decision rule for determining  
520 |a the number of sources. The simulation study shows that  
520 |a for the lognormal error case, they are superior to the  
520 |a traditional methods such as Bartlett's statistic or the  
520 |a rule of thumb methods, and for the normal error case, they  
520 |a are competitive with the best of the traditional methods.  
520 |a Many air pollution datasets typically consist of the  
520 |a measurements on fifty or sixty variables, and it is often  
520 |a too large to handle all at once. We develop two new  
520 |a algorithms, SPECIESA and SPECIESB, for choosing the  
520 |a species used in final model fitting. The simulation  
520 |a results show that these algorithms work very well in  
520 |a choosing the species generated by the sources not by the  
520 |a measurement errors. Estimating the source profiles and  
520 |a their contributions are our primary concerns in receptor  
520 |a modeling. We take the constrained nonlinear least squares  
520 |a approach to provide those estimates under as little  
520 |a assumptions as possible about the model parameters. The  
520 |a assumptions that we make are necessary to get the model 
520 |a identifiability. A set of algorithms, VERTEX, to find the  
520 |a least squares solution under different versions of the  
520 |a model is introduced. The resulting estimators are shown  
520 |a to be consistent and asymptotically normal under  
520 |a appropriate identifiability conditions. The proposed  
520 |a methods are applied to simulated data and real air  
520 |a pollution data. 
650 4 |a Major statistics. 
856 4 1 |u http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=736824721&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD  |t 0 
948 |a c:en 
999 |a MARS 
999 f f |s fc23b7d0-2bfc-3942-a452-9c54f91a7331  |i 85a9ee00-1d92-3566-b375-6dcd062fcdd7  |t 0 
952 f f |p noncirc  |a Texas A&M University  |b College Station  |c Cushing Memorial Library & Archives  |s cush tdrm  |d Cushing: Theses & Dissertations Microforms (Does not check out)  |t 0  |e 1997 Dissertation P3752  |h Other scheme  |i unmediated -- volume 
952 f f |a Texas A&M University  |b College Station  |c Electronic Resources  |s www_evans  |d Available Online  |t 0  |e 1997 Dissertation P3752  |h Other scheme 
998 f f |a 1997 Dissertation P3752  |t 0  |l Cushing: Theses & Dissertations Microforms (Does not check out) 
998 f f |a 1997 Dissertation P3752  |t 0  |l Available Online