Prediction of concentrations based on spectroscopic data is a common challenge. Least squares (LS) methods are an alternative to conventional regression methods. In this research potential and limits of Classical Least Squares (CLS), Generalized Least Squares (GLS), Extended Least Squares (ELS) and Multivariate Curve Resolution with Alternating Least Squares (MCR-ALS) are demonstrated in two study cases. The first part focuses on the reaction simulation, where the target compound compromises only a small portion of the mixture composition. Modifications of the Classical Least Squares method are superior in this case. The second part of the report deals with a practical case of a mixture of spent hydrogen sulfide scavengers when most compounds’ pure spectra are known. Aquaes solution of triazine is commonly used to remove toxic gas, hydrogen sulfide, in the oil and gas industry. Despite considerable research dedicated to the reaction of triazine with hydrogen sulfide, there needs to be more information about how to quantify the reactants and products of this reaction. This research is an attempt to quantify triazine and the products of its reaction with hydrogen sulfide in an aqueous solution using Raman spectroscopy. Raman spectroscopy combined with Least Squares methods gave promising results when tested on synthetic solutions containing components present in the spent scavenger solution. Afterward, Raman spectroscopy was used on real SUS samples and results were compared with conventional methods: 1H-NMR and GC-MS.