Lengyel, Paksy, Bánhidi, Héberger: COMPARISON OF SOME MULTIVARIATE METHODS FOR EVALUATION OF ENVIRONMENTAL DATA

COMPARISON OF SOME MULTIVARIATE METHODS FOR
EVALUATION OF ENVIRONMENTAL DATA

Attila Lengyel¹, László Paksy¹,
Olivér Bánhidi^1,2 and Károly Héberger³

¹University of Miskolc, Institute of Chemistry, Department of Analytical Chemistry,
Miskolc-Egyetemváros, H-3515, Miskolc, Hungary
²Metalcontrol Ltd. Miskolc, Vasgyári u 43, H-3540 Miskolc, Hungary
³Institute of Chemistry, Chemical Research Center, Hungarian Academy of Sciences,
H-1205, Budapest, Pusztaszeri u. 59/67.

Recently, methods that can handle several variables at once are very much favourable in the field of analytical chemistry, physical chemistry, metallurgy, mechanical engineering and other sciences. A lot of software offers comfortable and interactive possibilities for the applicants from the unskilled peoples to the high-qualified experts. Unfortunately, the easy computation has a hazard at the interpretation of the mathematical statistical results.

At this presentation the total and partial correlation are compared with the multiple linear regression and the Principal Component Analysis/Principal Component Regression (PCA/PCR).

For evaluation and qualify the state of environment by investigation of the ambient air using this methods, the PCA has two main advantages:

The data can be classified easily e.g. by place or time, etc.,
The strength of variables can be shown, so the significant variables can be selected easily.

Unfortunately, the results depend on the variable interval of the variables and the effective classification of data, consequently the interpretation of results requires knowledge either on the chemometric or on the basic science concerning the investigated problems.

References

[1] A. Lengyel, L. Paksy, O. Bánhidi: Chemometric Approach for Evaluating of Environmental Data, Proceedings of microCAD'99 Conference, Miskolc, Section A. 65-69. pp.

[2] A. Lengyel, L. Paksy, O. Bánhidi: Chemometric Evaluation of Environmental Data, IX Italian-Hungarian Symposium on Spectrochemistry, October 11-15, 1999, Siena, Italy

[3] A. Lengyel, L. Paksy, O. Bánhidi, K. Héberger: Chemometric methods for evaluation of environmental data, microCAD'2000 Conference, Miskolc, March 2-5. 2000.