In this thesis, several works, related to the use and development of different chemometric techniques applied to data treatment obtained from different sampling techniques (Transmittance, Attenuated Total Reflectance or Diffuse Reflectance) based on vibrational spectroscopy (Mid-IR and Near-IR ranges) are presented. In fact, their performance in the determination of different properties and quality parameters on a wide range of samples such as pharmaceuticals, food (juices, milk shakes, yoghurts, chocolates, and vinegars), agrochemicals products and tanned solid wastes used for removing hydrocarbons from polluted waters, are evaluated. Specifically, works included in this thesis expect to be a useful guide which tries to show up evolution from analytical methodologies requiring, in order to achieve certain selectivity, some chemical treatment on sample and for this fact they precise low data treatment; until that methodologies which requires no previous chemical treatment but, therefore, they precise relevant and complex mathematical treatment on instrumental data. For quantifying each property/parameter, a calibration model has been designed and optimized to get good results. In this sense, some works include analytical methods based on univariate calibration (zero order). On the other hand, analytical procedures using multivariate calibration (first order) models, which employ a series of mathematics algorithms on the different no-selective instrumental responses, are also described. Among different multivariate calibration techniques, Principal Component Analysis (PCA), Partial Least Squares Regression (PLSR), and Artificial Neural Networks (ANN), all of them with an appropriate mathematical pre-treatment, has been considered not only for classification but also to quantify properly the properties and parameters of different chemicals on the several samples under study. The most appropriate technique for each study has been used in order to asses the quality of the obtained results through the maximum predictive capabilities. Finally, a methodology based on the second-order data generation, combining multicommutation and vibrational spectroscopy, has been developed for the determination of acidity content in vinegar samples using multi-way decomposition method PARAFAC. Obtained results evidence that developed models are good, so established procedures, validated through their predictive capabilities, could be used to carry out a right quality control of the different samples considered in this Thesis.
