A new online computer program (BiDASys) for ordinary and uncertainty weighted least-squares linear regressions: Case studies from food chemistry

Abstract

A new computer program BiDASys (Bivariate Data Analysis System) is presented for the application of Ordinary and Uncertainty weighted least-squares linear regression models (OLR and UWLR) to experimental data from food chemistry. BiDASys has the following novel aspects: the statistical capability of detecting discordant outliers in bivariate data; new simulated critical values through Monte Carlo for the probability of no-correlation in multivariate samples (n=5-1000); and it is the only available program that can applied the UWLR model. The use of BiDASys is illustrated through three case studies where the relations 87Sr/86SrMust versus 87Sr/86Srsoil (from Glera-Prosecco, Italy), 87Sr/86SrGrape versus 87Sr/86SrWine,Soil (from Quebec, Canada), and 87Sr/86SrWines versus 87Sr/86SrRocks (from Tuscany-Basilicata, Italy) confirms that this isotopic ratio can be used to track the geographical origin of wine and one more case study from Guerrero (Mexico) concerning the influence of breastfeeding time on levels of organochlorine pesticides in human milk.