A combination of chemometrics methods and GC–MS for the classification of edible vegetable oils

https://doi.org/10.1016/j.chemolab.2016.03.028Get rights and content

Highlights

  • The fatty acid profiles of edible vegetable oils are obtained by GC–MS.

  • The SVM optimized by the GA is employed to construct the classification model.

  • The GA–SVM method is compared with other classification techniques.

Abstract

The authenticity of edible vegetable oils is a very important issue due to consumer health and commercial reasons. Gas chromatography–mass spectrometry (GC–MS) was applied to analyze the fatty acid composition of sixty six samples from six different kinds of edible vegetable oils. The fatty acid profiles of these edible vegetable oils were used to classify the type of edible oils. For improving the classification accuracy of vegetable oils with respect to type, the support vector machine (SVM) technique, optimized using the genetic algorithm (GA), was employed to construct the classification model. The effectiveness of the GA–SVM combination in classification was compared with that of other well-known strategies for classification, such as minimum distance classification (MDC) and linear discriminant analysis (LDA). In addition, the Kennard–Stone algorithm was used to select the representative training samples and compared with the random sampling method. The misclassification rates were 8.48% and 3.03% for training and test set, respectively, by the GA–SVM model using the linear kernel. Only one or two samples will be misclassified in the process of GA–SVM classification. The classification task based on fatty acid data can be successfully achieved by the GA–SVM technique combined with the Kennard–Stone algorithm. The results reveal that this strategy is of great promise in flexible and accurate classification of edible vegetable oils.

Introduction

Edible vegetable oils have been an indispensable ingredient of our diet in daily life since they contain a variety of essential fatty acids that are necessary for the human body by accelerating the absorption of fat-soluble vitamins [1], [2]. They can also provide the body with a direct source of energy. The most salient feature of vegetable oils is that their nutritional value is higher than animal oils. The long-term consumption of animal fats, comprising primarily saturated fatty acids, will increase the risk of hypertension and coronary heart disease [3], which makes edible vegetable oils especially attractive for people. The performances and qualities of different types of vegetable oils vary during homemade cooking and food production depending on their compositions [4]. Therefore, the authenticity of edible vegetable oils is a very important issue while considering commercial and consumer health reasons. In the present study, the continued need for improving the classification accuracy of edible vegetable oils is investigated.

Nowadays, routine methods of analysis of vegetable oils involve many instrumental analysis techniques, such as near and mid-infrared spectrometry [5], [6], fluorescence [7], chemiluminescence [8], chromatography [9], [10], nuclear magnetic resonance spectroscopy [11], and mass spectrometry [12]. Among these instrumental techniques mentioned above, gas chromatography–mass spectrometry (GC–MS) is frequently used to identify the vegetable oil type by analyzing their fatty acid composition [13], [14], [15], [16]. However, due to the complex composition of the vegetable oil, the resulting chromatograms may be formed by the overlapping of several analytical peaks. Furthermore, the chromatograms of different vegetable oils may be too similar to be distinguished directly. To further the improvement of previous methods, many researchers have proposed the spectroscopic techniques in combination with chemometrics methods as an alternative method that can be used for the discrimination of different categories and the detection of adulterants in oils [17], [18], [19]. The commonly used chemometrics methods include principal component analysis (PCA) [20], linear discriminant analysis (LDA) [21], and minimum distance classification (MDC). In the present study, a support vector machine (SVM) technique [22] was employed to construct the classification model with genetic algorithm (GA) [23] to get the optimized solution for edible vegetable oil classification.

SVM is a promising machine learning technique with comprehensive theoretical foundation. Because of the powerful ability in interpreting the linear or nonlinear relationships between the sample information and their properties, SVM has exhibited desirable generalization performance in numerous applications. GA simulates the Darwinian evolution of natural selection and genetic mechanism natural evolution. As a popular global stochastic optimization technique, GA has been successfully used for global search and optimization problem [24]. Here, GA is invoked to seek the optimal parameters for the classification model, including penalty constant and kernel parameter in a kernel transform of SVM. Using GA to get the optimal solution readily makes SVM an adaptive parameter-free method for edible vegetable oil identification, without any parameters to be adjusted. A synergetic optimization of the parameters also enables a flexible modeling approach for SVM according to the performance of the total model. The proposed strategy has been applied to the classification of 66 samples from six different kinds of edible vegetable oils. The effectiveness of the GA–SVM in classification ability was compared with that of other well-known strategies for classification, such as minimum distance classification (MDC) and linear discriminant analysis (LDA). To further improve the classification accuracy, the Kennard–Stone algorithm was employed for the selection of samples for each type of vegetable oil.

Section snippets

Chemical reagents and sample collection

All reagents used in the experiment were of analytical grade. Petroleum ether (boiling point 30–60 °C), methanol, and methylbenzene were provided by Sinopharm Chemical Reagent Co. Ltd (shanghai, China), potassium hydroxide from Shenghao chemical reagent Co. Ltd (Guangzhou, China), hydrochloric acid (12 M) from Kaixin chemical reagent Co. Ltd (Hengyang, China).

All solutions were prepared using ultrapure water, which was obtained through a Millipore Milli-Q water purification system (Billerica, MA,

Training and prediction set selection

The set of samples of edible oils was partitioned into two groups: a training set and a prediction set. Half of the samples were used for training purposes in classification studies, and the rest, constituting the prediction set, were used to evaluate the prediction capability of the classification model. There are several algorithms that can be used for the selection of samples for training and prediction sets. Among these algorithms, random sampling (RS) is widely used because of its

Characteristics of fatty acid profiles of various edible vegetable oils

Fig. 1 shows an overlay of the typical total ion chromatographic profiles of the derivatives of fatty acids for different edible vegetable oils obtained using GC–MS, covering a range of time from 700 s to 1400 s. The main composition of edible vegetable oils is triglyceride, which is composed of a variety of different fatty acids. It can be found that the types of fatty acids included in the six different kinds of vegetable oils were quiet similar. Typically, tetradecanoic acid (C14:0), palmitic

Conclusion

In the present study, MDC, LDA, and GA–SVM were employed to construct classification models for edible vegetable oils using a fatty acids data set of oil samples obtained from the GC-MS. It was verified that a combination of chemometrics methods and GC-MS could be a suitable tool for the classification of edible vegetable oils. Compared with the LDA and MDC models, the GA–SVM techniques combined with the Kennard–Stone algorithm can successfully classify the edible vegetable oils. The

Disclaimer

The material presented in this publication reflects the opinions of the authors, and not of their institutional affiliations.

Conflict of interest

There is no conflict of interest.

Acknowledgement

The work was financially supported by the National natural Science Foundation of China (Grant No. 21575131).

References (28)

Cited by (0)

View full text