Fruit juices are one of the most adulterated beverages, usually because of the addition of water, sugars, or less expensive fruit juices. This study presents a method based on Fourier transform infrared spectroscopy (FT-IR), in combination with machine learning methods, for the correct identification and quantification of adulterants in juices. Thus, three types of 100% squeezed juices (pineapple, orange, and apple) were evaluated and adulterated with grape juice at different percentages (5%, 10%, 15%, 20%, 30%, 40%, and 50%). The results of the exploratory data analysis revealed a clear clustering trend of the samples according to the type of juice analyzed. The supervised learning analysis, based on the development of models for the detection of adulteration, obtained significant results for all tested methods (i.e., support-vector machines or SVM), random forest or RF, and linear discriminant analysis or LDA) with an accuracy above 97% on the test set. Regarding quantification, the best results are obtained with the support vector regression and with partial least square regression showing an R² greater than 0.99 and a root mean square error (RMSE) less than 1.4 for the test set.

Honey is one of the most adulterated foods, usually through the addition of sweeteners or low-cost honeys. This study presents a method based on visible near infrared spectroscopy (Vis-NIRs), in combination with machine learning (ML) algorithms, for the correct identification and quantification of adulterants in honey. Honey samples from two botanical origins (orange blossom and sunflower) were evaluated and adulterated with low-cost honey in different percentages (5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, and 50%). The results of the exploratory analysis showed a tendency to group the samples according to botanical origin, as well as the presence of adulteration. A supervised analysis was performed to detect the presence of adulterations. The best performance with 100% accuracy was achieved by support vector machines (SVM) and random forests (RF). A regression study was also carried out to quantify the percentage of adulteration. The best result was obtained by support vector regression (SVR) with a coefficient of determination (R²) of 0.991 and a root mean squared error (RMSE) of 1.894. These results demonstrate the potential of combining ML with spectroscopic data as a method for the automated quality control of honey.

Fruit juice production is one of the most important sectors in the beverage industry, and its adulteration by adding cheaper juices is very common. This study presents a methodology based on the combination of machine learning models and near-infrared spectroscopy for the detection and quantification of juice-to-juice adulteration. We evaluated 100% squeezed apple, pineapple, and orange juices, which were adulterated with grape juice at different percentages (5%, 10%, 15%, 20%, 30%, 40%, and 50%). The spectroscopic data have been combined with different machine learning tools to develop predictive models for the control of the juice quality. The use of non-supervised techniques, specifically model-based clustering, revealed a grouping trend of the samples depending on the type of juice. The use of supervised techniques such as random forest and linear discriminant analysis models has allowed for the detection of the adulterated samples with an accuracy of 98% in the test set. In addition, a Boruta algorithm was applied which selected 89 variables as significant for adulterant quantification, and support vector regression achieved a regression coefficient of 0.989 and a root mean squared error of 1.683 in the test set. These results show the suitability of the machine learning tools combined with spectroscopic data as a screening method for the quality control of fruit juices. In addition, a prototype application has been developed to share the models with other users and facilitate the detection and quantification of adulteration in juices.

Interpretation of data from fire debris is considered as one of the most challenging steps in fire investigation. Forensic analysts are tasked to identify the presence or absence of ignitable liquid residues (ILRs) which may indicate whether a fire was started deliberately. So far, data analysis is subjected to human interpretation following the American Society for Testing and Materials’ guidelines (ASTM E1618) based on gas chromatography–mass spectrometry data. However, different factors such as interfering pyrolysis compounds may hinder the interpretation of data. Some substrates release compounds that are in the range of common ignitable liquids, which interferes with accurate determination of ILRs. The aim of the current research is to investigate whether headspace–mass spectroscopy electronic nose (HS-MS eNose) combined with pattern recognition can be used to classify different ILRs from fire debris samples that contain a complex matrix (petroleum-based substrates or synthetic fibers carpet) that can strongly interfere with their identification. Six different substrates—four petroleum-derived substrates (vinyl, linoleum, polyester, and polyamide carpet), as well as two different materials for comparison purposes (cotton and cork) were used to investigate background interferences. Gasoline, diesel, ethanol, and charcoal starter with kerosene were used as ignitable liquids. In addition, fire debris samples were taken after different elapsed times. A total of 360 fire debris samples were analyzed. The obtained total ion mass spectrum was combined with unsupervised exploratory techniques such as hierarchical cluster analysis (HCA) as well as supervised linear discriminant analysis (LDA). The results from HCA show a strong tendency to group the samples according to the ILs and substrate used, and LDA allowed for a full identification and discrimination of every ILR regardless of the substrate.

Sherry wine vinegar is a Spanish gourmet product under Protected Designation of Origin (PDO). Before a vinegar can be labeled as Sherry vinegar, the product must meet certain requirements as established by its PDO, which, in this case, means that it has been produced following the traditional solera and criadera ageing system. The quality of the vinegar is determined by many factors such as the raw material, the acetification process or the aging system. For this reason, mainly producers, but also consumers, would benefit from the employment of effective analytical tools that allow precisely determining the origin and quality of vinegar. In the present study, a total of 48 Sherry vinegar samples manufactured from three different starting wines (Palomino Fino, Moscatel, and Pedro Ximénez wine) were analyzed by Fourier-transform infrared (FT-IR) spectroscopy. The spectroscopic data were combined with unsupervised exploratory techniques such as hierarchical cluster analysis (HCA) and principal component analysis (PCA), as well as other nonparametric supervised techniques, namely, support vector machine (SVM) and random forest (RF), for the characterization of the samples. The HCA and PCA results present a clear grouping trend of the vinegar samples according to their raw materials. SVM in combination with leave-one-out cross-validation (LOOCV) successfully classified 100% of the samples, according to the type of wine used for their production. The RF method allowed selecting the most important variables to develop the characteristic fingerprint (“spectralprint”) of the vinegar samples according to their starting wine. Furthermore, the RF model reached 100% accuracy for both LOOCV and out-of-bag (OOB) sets.

Fruit juices are one of the most widely consumed beverages worldwide, and their production is subject to strict regulations. Therefore, this study presents a methodology based on the use of headspace–gas chromatography–ion mobility spectrometry (HS-GC-IMS) in combination with machine-learning algorithms for the characterization juices of different raw material (orange, pineapple, or apple and grape). For this purpose, the ion mobility sum spectrum (IMSS) was used. First, an optimization of the most important conditions in generating the HS was carried out using a Box–Behnken design coupled with a response surface methodology. The following factors were studied: temperature, time, and sample volume. The optimum values were 46.3 °C, 5 min, and 750 µL, respectively. Once the conditions were optimized, 76 samples of the different types of juices were analyzed and the IMSS was combined with different machine-learning algorithms for its characterization. The exploratory analysis by hierarchical cluster analysis (HCA) and principal component analysis (PCA) revealed a clear tendency to group the samples according to the type of fruit juice and, to a lesser extent, the commercial brand. The combination of IMSS with supervised classification techniques reported an excellent result with 100% accuracy on the test set for support vector machines (SVM) and random forest (RF) models regarding the specific fruit used. Nevertheless, all the models have proven to be an effective alternative for characterizing and classifying the different types of juices.