Premium
Raw data pre‐processing in the protozoa and metazoa identification by image analysis and multivariate statistical techniques
Author(s) -
Ginoris Y. P.,
Amaral A. L.,
Nicolau A.,
Coelho M. A. Z.,
Ferreira E. C.
Publication year - 2007
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.1054
Subject(s) - normalization (sociology) , linear discriminant analysis , protozoa , population , artificial intelligence , computer science , pattern recognition (psychology) , biology , biological system , microbiology and biotechnology , demography , sociology , anthropology
Different protozoa and metazoa populations develop in the activated sludge wastewater treatment processes and are highly dependent on the operating conditions. In the current work the protozoa and metazoa groups and species most frequent in wastewater treatment plants were studied, mainly the flagellate, sarcodine, and ciliate protozoa as well as the rotifer, gastrotrichia , and oligotrichia metazoa. The work is centered on the survey of the wastewater treatment plant conditions by protozoa and metazoa population using image analysis, discriminant analysis (DA), and neural networks (NNs) techniques, and its main objective was set on the evaluation of the importance of raw data pre‐processing techniques in the final results. The main pre‐processing techniques herein studied were the raw parameters reduction set by a joint cross‐correlation and decision trees (DTs) procedure and two data normalization techniques: logarithmic normalization and standard deviation normalization. Regarding the parameters reduction methodology, the use of a joint DTs and correlation analysis (CA) procedure resulted in 28 and 30% reductions in terms of the initial parameters set for the stalked and non‐stalked microorganisms, respectively. Consequently, the use of the reduced parameters set has proven to be a suitable starting point for both the DA and NNs methodologies, although for the DA an initial logarithmic normalization step is advisable. For the NNs analysis a standard deviation normalization procedure could be considered for the non‐stalked microorganisms regarding the operating parameters assessment. Copyright © 2007 John Wiley & Sons, Ltd.