Combining wavelength importance ranking to the random forest classifier to analyze multiclass spectral data.

Near Infrared (NIR) is a type of vibrational spectroscopy widely used in different areas to characterize substances. NIR datasets are comprised of absorbance measures on a range of wavelengths (λ). Typically noisy and correlated, the use of such datasets tend to compromise the performance of several statistical techniques; one way to overcome that is to select portions of the spectra in which wavelengths are more informative. In this paper we investigate the performance of the Random Forest (RF) classifier associated with several wavelength importance ranking approaches on the task of classifying product samples into categories, such as quality levels or authenticity. Our propositions are tested using six NIR datasets comprised of two or more classes of food and pharmaceutical products, as well as illegal drugs. Our proposed classification model, an integration of the χ2 ranking score and the RF classifier, substantially reduced the number of wavelengths in the dataset, while increasing the classification accuracy when compared to the use of complete datasets. Our propositions also presented good performance when compared to competing methods available in the literature.

Wavenumber selection based on Singular Value Decomposition for sample classification.

The dissemination of falsified medicines is a public health risk. Techniques such as attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy are commonly adopted for fraudulent drug detection. However, the spectrum generated by the ATR-FTIR typically results in hundreds of wavenumbers, reducing the performance of classification methods aimed at discriminating between authentic and falsified medicines. This article proposes a novel method for selecting a reduced size subset of wavenumbers that improves the classifier performance. The singular value decomposition SVD is used to generate a wavenumber importance index. An iterative process creates k-nearest neighbor (KNN) models by adding the wavenumbers in a decreasing order according to the importance index. Wavenumbers that increase classification accuracy are selected. When applied to Cialis® ATR-FTIR data, the proposed approach retained average 0.51% of the original wavenumbers with 100% accurate classifications; as for the Viagra® data set, the method yielded perfect classifications retaining average 0.17% of the original wavenumbers.