Multivariate linear regression with variable selection by a successive projections algorithm applied to the analysis of anodic stripping voltammetry data

• MLR aided by variable selection provided excellent quantitative predictions.
• The electrochemical processes occurring in ASV can be outlined simultaneously.
• Peak alignment gives better prediction results and simpler models.

Multivariate linear regression aided by a successive projections algorithm (SPA-MLR) was applied in the evaluation of anodic stripping voltammetry data obtained in the simultaneous determination of metals under conditions where there were significant complications due to interference processes such as the formation of intermetallic compounds and overlapping peaks. Using simulated data, modeled from complex interactions experimentally observed in samples containing Cu and Zn, as well as Co and Zn, it was demonstrated that SPA-MLR selected variables that allow chemical interpretation. This feature was used to make inferences about the underlying electrochemical processes during the simultaneous determination of four metals (Cu, Pb, Cd, and Co) in a concentration range where all responses were complicated by interference processes (10-100 ng mL−1). Additionally, the analytical performances of MLR models for quantitative predictions were excellent despite the complexity of the system under study.