Repositori URV

Identificador: imarina:9364440

Handle: https://hdl.handle.net/20.500.11797/imarina9364440

Autors:
Ezenarro, J.Schorn-García, D.Busto, O.Boqué, R.

Resum:
This paper introduces the ProSpecTool, a MATLAB toolbox for automated selection of preprocessing methods for obtaining optimal PLS regression models in vibrational spectroscopy. Trial-and-error approaches for preprocessing can be time-consuming, and the success of the process relies on the experience of analysts. The ProSpecTool addresses this challenge by using objective criteria analogous to expert judgment to filter and iterate preprocessing methods based on raw data properties. The toolbox quantifies noise, identifies multiplicative and additive scatter-effects, and selects preprocessing algorithms to correct them. Results demonstrate that the ProSpecTool can produce models that resemble those proposed by experienced analysts based on trial-and-error in terms of performance and robustness, making it a valuable exploratory tool for vibrational spectroscopy practitioners.