Jorge Costa Pereira1, Paweł K Zarzycki2. 1. University of Coimbra, Centro de Química de Coimbra (CQC), Department of Chemistry, Rua Larga, Coimbra P-3004-535, Portugal. 2. Koszalin University of Technology, Faculty of Civil Engineering, Environmental and Geodetic Sciences, Department of Environmental Technologies and Bioanalytics, Sniadeckich 2, Koszalin 75-453, Poland.
Abstract
BACKGROUND: Sampling (collecting) and probing (testing, measuring) are very common tasks in the analytical field, where we need to characterize a given system and complex samples. In this action, we try to ensemble maximal information related with the system under a given study, and, frequently, we may end an inefficient analytical situation. OBJECTIVE: The best way to avoid "oversampling" and "overprobing" is to evaluate the number of factors and objects that may be present in a current data set. METHODS: Suggested methodology in data analysis is mainly related with principal component analysis and principal object analysis. All used simulations and other controlled situations were here used to demonstrate how to retrieve the number of factors and objects present in a given data set and allow to supervise all sampling and probing process. RESULTS AND CONCLUSIONS: In this work, we explain and suggest how to use eigenvalue decomposition to access the actual number of factors and object contributions. A large pool of datasets were tested in order to assess the number of relevant features present in each dataset. HIGHLIGHTS: Proposed numerical approach was designed to supervise and help in sampling and probing process for the efficient analysis of complex systems such as those involving food and environmental samples.
BACKGROUND: Sampling (collecting) and probing (testing, measuring) are very common tasks in the analytical field, where we need to characterize a given system and complex samples. In this action, we try to ensemble maximal information related with the system under a given study, and, frequently, we may end an inefficient analytical situation. OBJECTIVE: The best way to avoid "oversampling" and "overprobing" is to evaluate the number of factors and objects that may be present in a current data set. METHODS: Suggested methodology in data analysis is mainly related with principal component analysis and principal object analysis. All used simulations and other controlled situations were here used to demonstrate how to retrieve the number of factors and objects present in a given data set and allow to supervise all sampling and probing process. RESULTS AND CONCLUSIONS: In this work, we explain and suggest how to use eigenvalue decomposition to access the actual number of factors and object contributions. A large pool of datasets were tested in order to assess the number of relevant features present in each dataset. HIGHLIGHTS: Proposed numerical approach was designed to supervise and help in sampling and probing process for the efficient analysis of complex systems such as those involving food and environmental samples.