Partial Least Squares Discriminant Analysis Method Research Articles

Geographical Origin Identification of Tequila Based on Multielement and Stable Isotopes

To explore the regional characteristics of multielement and stable isotopes in tequila and the viability of tracing the geographical origin, both inductively coupled plasma mass spectrometry (ICP-MS) and stable isotope ratio mass spectrometry (IRMS) were applied in this study to determine the multielement content and stable isotope ratios of tequila. Then, principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) as chemometrics were conducted in combination to establish a model for determining the geographical origin of Mexican tequila. According to the analytical results, the content of 22 elements varied, as did the ranges of stable isotope δ13C and δ18O values for the tequila obtained from the 4 producing areas. Besides, there were regional characteristics shown to some extent. PCA and PLS-DA methods can be adopted to identify the geographical origin of tequila in Mexico. According to the discriminant verification of 16 samples, the accuracy of prediction reached 93.75%. It has been demonstrated that it is possible to distinguish tequila of different geographical origins in Mexico by combining the determination of the multielement content and stable isotope ratios with chemometric analysis, which lays a foundation for tracing the geographical origin of tequila.

Raman spectroscopy and surface-enhanced Raman spectroscopy (SERS) spectra of salivary glands carcinoma, tumor and healthy tissues and their homogenates analyzed by chemometry: Towards development of the novel tool for clinical diagnosis

In this study, two approaches to salivary glands studies are presented: Raman imaging (RI) of tissue cross-section and surface-enhanced Raman spectroscopy (SERS) of tissue homogenates prepared according to elaborated protocol. Collected and analyzed data demonstrate the significant potential of SERS combined with multivariate analysis for distinguishing carcinoma or tumor from the normal salivary gland tissues as a rapid, label-free tool in cancer detection in oncological diagnostics. Raman imaging allows a detailed analysis of the cell wall's chemical composition; thus, the compound's distribution can be semi-quantitatively analyzed, while SERS of tissue homogenates allow for detailed analysis of all moieties forming these tissues. In this sense, SERS is more sensitive and reliable to study any changes in the area of infected tissues. Principal component analysis (PCA), as an unsupervised pattern recognition method, was used to identify the differences in the SERS salivary glands homogenates. The partial least squares-discriminant analysis (PLS-DA), the supervised pattern classification technique, was also used to strengthen further the computed model based on the latent variables in the SERS spectra. Moreover, the chemometric quantification of obtained data was analyzed using principal component regression (PCR) multivariate calibration. The presented data prove that the PCA algorithm allows for 91% in seven following components and the determination between healthy and tumor salivary gland homogenates. The PCR and PLS-DA methods predict 90% and 95% of the variance between the studied groups (in 6 components and 4 factors, respectively). Moreover, according to calculated RMSEC (RMSEP), R2C (R2P) values and correlation accuracy (based on the ROC curve), the PLS-DA model fits better for the studied data. Thus, SERS methods combined with PLS-DA analysis can be used to differentiate healthy, neoplastic, and mixed tissues as a competitive tool in relation to the commonly used method of histopathological staining of tumor tissue.

A rapid and non-destructive ATR-FTIR spectroscopy method supported by chemometrics for discriminating between facial creams and the classification into herbal and non-herbal brands

Facial creams are considered to be essential beauty items and are used by both females and males on an everyday basis. These can be encountered as an evidentiary material in criminal investigations, particularly in cases related to sexual and physical assaults against women. These are found in trace amounts and therefore their analysis is difficult and also, it must be through non-destructive methods. In the present work ATR-FTIR spectroscopy was employed for the discrimination of 57 samples of face creams out of which 31 were non-herbal and 26 were from herbal category. Visual analysis of the obtained Spectra was done for discrimination purposes but the method was prone to human error and laborious too. The spectroscopic results were analyzed with PCA (Principal Component Analysis) and PLS-DA (Partial least square discriminant analysis) methods. A segregation of samples was seen in the PCA plots to some extent. The class separation and prediction of the samples was performed using PLS-DA method. A good classification was achieved between herbal and non-herbal samples using PLS-DA method. Further, validation of the model was also performed by testing 10 unknown samples.

Determination of the Effective Economic and/or Demographic Indicators in Classification of European Union Member and Candidate Countries Using Partial Least Squares Discriminant Analysis

Partial Least Squares Discriminant Analysis (PLSDA) is a statistical method for classification and consists of a classical Partial Least Squares Regression in which the dependent variable is a categorical one expressing the class membership of each observation. The aim of this study is both analyzing the performance of PLSDA method in classifying 28 European Union (EU) member countries and 7 candidate countries (Albania, Montenegro, Serbia, Macedonia FYR, Turkey moreover including potential candidates Bosnia and Herzegovina and Kosova) correctly to their pre-defined classes (candidate or member) and determining the economic and/or demographic indicators, which are effective in classifying, by using the data set obtained from database of the World Bank.

Comparative investigation of raw and processed products of Gardeniae Fructus and Gardenia jasminoides var. radicans using HPLC coupled with chemometric methods.

As a commonly used traditional Chinese medicine (TCM), Gardeniae Fructus (GF) and its processed products, GF (stir-baked) and GF Praeparatus, have important medicinal value in clinical practice. Gardenia jasminoides var. radicans (GJVR) is a variant of GF, and because of the naming GJVR is often confused in the clinic with GF, resulting in medical misprescriptions. To distinguish GF and GJVR and study the changes before and after processing, the fingerprints of GF and GJVR are presented using HPLC, followed by hierarchical cluster analysis (HCA), principal component analysis (PCA), and partial least squares-discriminant analysis (PLS-DA). GF has purging and choleretic effects, and in this study, we determined the content of main active ingredients to preliminarily assess the GF and GJVR quality from the perspective of material basis. For PCA score plot, the samples fell into six clusters, the cross-validity Q2 (cum) = 0.842 and the cumulative contribution rate R2 x (cum) = 0.988, indicating that the model has a good precision. The results were then corroborated by HCA and PLS-DA method, showing that this methodology can distinguish GF and GJVR and can be used for the comparison of raw and two processed products. According to the model established by PLS-DA, eight components were identified as the most significant variables for discrimination. The results obtained by multiple model methods are consistent and verified by each other, providing a scientific reference for further clarification of the medicinal properties of GF and GJVR.

Fractional order modeling and recognition of nitrogen content level of rubber tree foliage

The Nondestructive estimation method of nitrogen content level of rubber tree foliage was investigated utilizing near infrared (NIR) spectroscopy and Grünwald-Letnikov fractional calculus. Four models, including partial least squares discriminant analysis (PLS-DA), support vector machine (SVM), extreme learning machine (ELM) and convolutional neural networks (CNN) are applied to construct the nitrogen estimation model. The results show that models established by 0.6-order or 1.6-order spectra achieved better performance than models with integer-order spectra. Afterward, the successive projections algorithm (SPA) is applied to reduce the number of variables, which is critical for developing portable nitrogen-level detector devices for rubber trees. The PLS-DA method achieved the best performance with an optimal recognition rate (97.73%) using the 1.6-order spectra. The results suggest that nitrogen content of rubber trees could be reliably estimated by fractional calculus processed NIR spectra. The method proposed here has a wide range of applicability and can provide more useful information for NIR spectral analysis in agriculture as well as other fields.

Identification of rice flour types with near-infrared spectroscopy associated with PLS-DA and SVM methods

One of the most important problems associated with the rice industry is the authenticity, mainly the identification of varieties by providing a reliable, fast, yet accurate method. To overcome these limitations, the development of fast and non-destructive methodologies for different rice type classification is, nowadays, a huge challenge for producers. The near-infrared (NIR) spectroscopy associated to principal component analysis (PCA), partial least squares-discriminant analysis (PLS-DA), and support vector machines (SVM) for discrimination and classification of rice varieties (Indica and Japonica) were explored after different spectra processing steps such as multiplicative scatter correction (MSC), first derivative and second derivative. The SVM model developed after the MSC processing procedure, showed a significant fitting accuracy (97%), cross-validation (93%) and prediction (91%). These data support the robustness of the model for efficient rice types classification. In terms of spectral analysis, the major differences between both rice types are present at range 7476–7095 cm−1, 7046 cm−1 and 4264–4153 cm−1, which can be used for its discrimination. This study showed that NIR spectroscopy associated to PLS-DA and SVM techniques allowed an efficient discrimination of rice samples, being considered as a suitable strategy for a competent system for fully automated classification and sorting of rice types grouping with a high level of accuracy, representing a valuable approach for discrimination and anti-fraud procedure for food control as well as in terms of security issues of any product.

Predictors of fatal outcomes in pedestrian accidents in Tabriz Metropolis of Iran: Application of PLS-DA method

Objectives: Road traffic deaths in walking pedestrians are a global public health problem. Considering that in Iran pedestrians have a high proportion of deaths caused by traffic accidents, the objective of the present study was to investigate mortality rate and related factors of fatal injury in pedestrian crashes in Tabriz Metropolis of Iran as the largest and most populous city of the northwest of Iran.Methods: The design of this study is case-control based on police and Forensic Medicine Organization data. All registered fatal pedestrian crashes in Tabriz Metropolis from 2014 to 2015 (146 cases) were included in the study as the case group. Also, 292 pedestrians (the ratio of cases to controls was 1:2) with non-fatal crashes were considered as the control group. Due to high dimensional data and multicollinearity issue, Partial least squares discriminant analysis (PLS-DA) was used for data analysis. Importance of the variables was determined by the VIP (Variable Importance in the Projection) index. Performance of the model was assessed by using training and test set validation method. The area under the ROC curve (AUC) and classification error rates were calculated for the test set. R software version 3.5.1 (mixOmcs packages) was used for data analysis.Results: According to the results of PLS-DA, the most important variables related to fatal outcome in pedestrian crashes with VIP > 1 were: pedestrian age (positive effect), type of vehicle (light machinery with a negative effect), kind of vehicle plate (private plate with a negative effect), season of accident occurrence (winter season with a positive effect), type of driver’s licenses (Class A with a positive effect), pedestrian gender (male with a positive effect) and Fault of Pedestrian (At-fault with a positive effect). The overall accuracy for the fitted model and AUC were 0.77 and 0.79, respectively.Conclusions: The results show that predictors of a fatal outcome in pedestrian accidents in Tabriz can be attributed to the pedestrian characteristics (which notably account for differences in vulnerability in case of an accident), the car and driver features, and weather (which may all notably influence the amount of energy involved in the collision, through the car mass, speed, and conditions delaying the braking response or reducing the braking effectiveness). Regarding the statistical method, the PLS-DA is a powerful method which can be used to analyze high dimension data with multicollinearity issue.

Locality preserving partial least squares discriminant analysis for face recognition

We propose a locality preserving partial least squares discriminant analysis (LPPLSDA) which adds a locality preserving feature to the conventional partial least squares discriminant analysis(PLS-DA). The locality preserving feature captures the within group structural information via a similarity graph. The ability of LPPLS-DA to capture local structures allows it to be better suited for face recognition. We evaluate the performance of our proposed method on several benchmarked face databases which offer different levels of complexity in terms of sample size as well as image acquisition conditions. The experimental results indicate that, for each database used, the proposed method consistently outperformed the conventional PLS-DA method.

Metabolomic biomarkers in cervicovaginal fluid for detecting endometrial cancer through nuclear magnetic resonance spectroscopy

Endometrial cancer (EC) is one of the most common gynecologic neoplasms in developed countries but lacks screening biomarkers. We aim to identify and validate metabolomic biomarkers in cervicovaginal fluid (CVF) for detecting EC through nuclear magnetic resonance (NMR) spectroscopy. We screened 100 women with suspicion of EC and benign gynecological conditions, and randomized them into the training and independent testing datasets using a 5:1 study design. CVF samples were analyzed using a 600-MHz NMR spectrometer equipped with a cryoprobe. Four machine learning algorithms-support vector machine (SVM), partial least squares discriminant analysis (PLS-DA), random forest (RF), and logistic regression (LR), were applied to develop the model for identifying metabolomic biomarkers in cervicovaginal fluid for EC detection. A total of 54 women were eligible for the final analysis, with 21 EC and 33 non-EC. From 29 identified metabolites in cervicovaginal fluid samples, the top-ranking metabolites chosen through SVM, RF and PLS-DA which existed in independent metabolic pathways, i.e. phosphocholine, malate, and asparagine, were selected to build the prediction model. The SVM, PLS-DA, RF, and LR methods all yielded area under the curve values between 0.88 and 0.92 in the training dataset. In the testing dataset, the SVM and RF methods yielded the highest accuracy of 0.78 and the specificity of 0.75 and 0.80, respectively. Phosphocholine, asparagine, and malate from cervicovaginal fluid, which were identified and independently validated through models built using machine learning algorithms, are promising metabolomic biomarkers for the detection of EC using NMR spectroscopy.

Authentication of organic Lampung robusta ground roasted coffee by UV-visible spectroscopy and PLS-DA method

In this research, a potential application of UV-visible (UV-Vis) spectroscopy combined with partial least squares-discriminant analysis (PLS-DA) method to discriminate Lampung robusta coffee with different fertilizer treatment was evaluated. The fully red ripened coffee beans were selectively harvested by hand from coffee plantation located in Lampung Barat of Lampung province from two different fertilizer treatments: chemically fertilized and organically fertilized. A number of 200 ground roasted coffee samples of each treatment (1 gram of each samples) was used as samples, respectively. The all coffee samples were extracted using hot distilled water. The aqueous coffee samples were pipetted into 10 mm of cuvette and the spectral data was obtained using a UV-Vis spectrometer in the range of 190-1100 nm. Principal component analysis (PCA) and PLS-DA method was used as unsupervised and supervised classification methods to discriminate the organic and non-organic coffee. The results showed that using the first two principal components (PCs), a clear separation between organic and non-organic coffee samples was achieved using modified spectral data in the range of 230-450 nm. The classification of organic and non-organic coffee using PLS-DA method resulted in high accuracy both for calibration and prediction steps. The overall result showed that UV-visible spectroscopy combined with PLS-DA method could be used as a low-cost, relative fast and green method to discriminate between organic and non-organic Lampung robusta ground roasted coffee.

Attenuated Total Reflection-Fourier Transform Infrared Spectroscopy (ATR-FTIR) Combined with Chemometrics Methods for the Classification of Lingzhi Species.

Due to the existence of Lingzhi adulteration, there is a growing demand for species classification of medicinal mushrooms by various techniques. The objective of this study was to explore a rapid and reliable way to distinguish between different Lingzhi species and compare the influence of data pretreatment methods on the recognition results. To this end, 120 fresh fruiting bodies of Lingzhi were collected, and all of them were analyzed by attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR). Random forest (RF), support vector machine (SVM) and partial least squares discriminant analysis (PLS-DA) classification models were established for raw and pretreated second derivative (SD) spectral matrices to authenticate different Lingzhi species. The results of multivariate statistical analysis indicated that the SD preprocessing method displayed a higher classification ability, which may be attributed to the analysis of powder samples that requires removal of overlapping peaks and baseline shifts. Compared with RF, the results of the SVM and PLS-DA methods were more satisfying, and their accuracies for the test set were both 100%. Among SVM and PLS-DA, the training set and test set accuracy of PLS-DA were both 100%. In conclusion, ATR-FTIR spectroscopy data pretreated by SD combined with PLS-DA is a simple, rapid, non-destructive and relatively inexpensive method to discriminate between mushroom species and provide a good reference to quality assessment.

The authentication of peaberry and civet ground roasted robusta coffee using UV-visible spectroscopy and PLS-DA method with two different particle sizes

The objective of this research was to investigate the influence of particle size of ground roasted coffee samples in the authentication of peaberry and civet coffee using UV-visible spectroscopy combined with partial least squares-discriminant analysis (PLS-DA) method. For this purpose, we provide a total of 80 samples of peaberry and civet coffee with two different particle size of 1680 micrometer (mesh 12) and 297 micrometer (mesh 50). The all coffee samples were extracted using hot distilled water (90-98°C). UV-visible spectral data of the all coffee samples were acquired using a UV-visible spectrometer in transmittance mode. The result showed that the different absorbance spectra were observed for the different particle sizes of peaberry and civet coffee samples. The absorbance of samples having particle size of 297 micrometer (mesh 50) is higher than that of particle size of 1680 micrometer (mesh 12). The calibration model was developed using PLS-DA method for samples having same particle size and resulted in acceptable result with high coefficient of determination (R2) (R2=0.99 both for mesh 12 and 50, respectively), low root mean square error of calibration (RMSEC), and high residual prediction deviation (RPD). However, the performance of the prediction results was very low for samples having different particle size with high bias and high root mean square error of prediction (RMSEP).

The Feasibility of Geographical Origin Discrimination of Lampung Robusta Coffee Using UV-Visible Spectroscopy and Chemometric Methods

In this study, UV-visible spectroscopy and partial least squares-discriminant analysis (PLS-DA) method was used to discriminate the origin of Lampung robusta coffee. Total 40 Lampung robusta coffee samples with two origins (Lampung Barat and Tanggamus) were used. Each samples has 1 gram weight and was extracted using hot distilled water. UV-visible spectral data of the all aqueous coffee samples were obtained using a UV-visible spectrometer in transmittance mode in the range of 190-1100 nm. The performance of the developed PLS-DA model was evaluated based on coefficient of determination (R2), root mean square error of calibration (RMSEC), and residual prediction to deviation (RPD). Using Savitzky-Golay 1st derivative and moving average spectra, PLS-DA model was developed with the following results: R2 = 0.97 for calibration and R2 = 0.88 for validation, RMSEC = 0.089421 and RMSECV=0.178384. The performance of prediction was quite good with RMSEP = 0.215303 and RPD = 2.83. This results show that UV-visible spectroscopy and PLS-DA method is a promising analytical method to discriminate geographical origin of Lampung robusta coffee.

Raman spectroscopic detection of high-grade cervical cytology: Using morphologically normal appearing cells

This study aims to detect high grade squamous intraepithelial cells (HSIL) by investigating HSIL associated biochemical changes in morphologically normal appearing intermediate and superficial cells using Raman spectroscopy. Raman spectra (n = 755) were measured from intermediate and superficial cells from negative cytology ThinPrep specimens (n = 18) and from morphologically normal appearing intermediate and superficial cells from HSIL cytology ThinPrep specimens (n = 17). The Raman data was subjected to multivariate algorithms including the standard principal component analysis (PCA)-linear discriminant analysis (LDA) and partial least squares discriminant analysis (PLS-DA) together with random subsets cross-validation for discriminating negative cytology from HSIL. The PCA-LDA method yielded sensitivities of 74.9%, 72.8%, and 75.6% and specificities of 89.9%, 81.9%, and 84.5%, for HSIL diagnosis based on the dataset obtained from intermediate, superficial and mixed intermediate/superficial cells, respectively. The PLS-DA method provided improved sensitivities of 95.5%, 95.2% and 96.1% and specificities of 92.7%, 94.7% and 93.5% compared to the PCA-LDA method. The results demonstrate that the biochemical signatures of morphologically normal appearing cells can be used to discriminate between negative and HSIL cytology. In addition, it was found that mixed intermediate and superficial cells could be used for HSIL diagnosis as the biochemical differences between negative and HSIL cytology were greater than the biochemical differences between intermediate and superficial cell types.

Variable selection for multivariate classification aiming to detect individual adulterants and their blends in grape nectars

During the quality inspection control of fruit beverages, some types of adulterations can be detected, such as the addition or substitution with less expensive fruits. To determine whether grape nectars were adulterated by substitution with apple or cashew juice or by a mixture of both, a methodology based on attenuated total reflectance Fourier transform mid infrared spectroscopy (ATR-FTIR) and multivariate classification methods was proposed. Partial least squares discriminant analysis (PLS-DA) and soft independent modeling of class analogy (SIMCA) models were developed as multi-class methods (classes unadulterated, adulterated with cashew and adulterated with apple) with the full-spectra. PLS-DA presented better performance parameters than SIMCA in the classification of samples with just one adulterant, while poor results were achieved for samples with blends of two adulterants when using both classification methods. Three variable selection methods were tested in order to improve the effectiveness of the classification models: interval partial least squares (iPLS), variable importance in projection scores (VIP scores) and a genetic algorithm (GA). Variable selection methods improved the performance parameters for the SIMCA and PLS-DA methods when they were used to predict samples with only one adulterant. Only PLS-DA coupled with iPLS was able to classify samples with blends of two adulterants, providing sensitivity values between 100% and 83% at 100% specificity for the three studied classes.

Protective effects of timosaponin BII on oxidative stress damage in PC12 cells based on metabolomics.

Peroxide and oxygen free radicals are some of the causes of oxidative stress in brain tissue, and could lead to the change of brain structure and function. In addition, oxidative damage is one of the most important causes of the aging of the vast majority of tissues. The aim of this study is to investigate the protective effect of timosaponin BII on oxidative stress damage of PC12 induced by H2 O2 using metabolomics based on the UHPLC-Q-TOF-MS technique. Partial least-squares discriminant analysis method was used to identify 35 metabolites as decisive marker compounds in a preliminary interpretation of the mechanism of the antioxidative effect of timosaponin BII. The majority of these metabolites are involved in the glutathione metabolism, amino acid metabolism, sphingolipid and glycerophospholipid metabolism. Our results suggest that timosaponin BII demonstrates systematic antioxidant effects in the PC12 oxidative damage cell model via the regulation of multiple metabolic pathways. These findings provide insight into the pathophysiological mechanisms underlying oxidative stress damage and suggest innovative and effective treatments for this disorder, providing a reliable basis for the development of novel therapeutic target in timosaponin BII treatment of oxidative stress.

Identification of Adulterated and Non-adulterated Norwegian Salmon Using FTIR and an Improved PLS-DA Method

Norwegian salmon is often adulterated with Heilongjiang salmon at local fish markets. To promote fair price competition at fish markets and protect consumer rights, we developed a quick and accurate identification method to distinguish adulterated and non-adulterated Norwegian salmon using a Fourier transform infrared spectroscopy (FTIR). In this study, Norwegian and Heilongjiang salmon could be readily distinguished using partial least squares discriminant analysis (PLS-DA), but it failed to detect the accurate level of 20 to 80% of adulterated Norwegian salmon samples. In order to improve the PLS-DA model, several pre-processing methods, including standard normal variate (SNV), multiplicative scatter correction (MSC), and normalization, were used to evaluate individually to select the most appropriate correction method. Characteristics of the spectra within the waveband range covering 450 to 4000 cm−1 were also analyzed to determine the optimum sub-waveband range to improve the accuracy of the model. The results of the study showed that using FTIR and the improved PLS-DA model established in this study, the adulterated and non-adulterated Norwegian salmon could be completely distinguished. The accuracy of the adulteration level and the prediction accuracy of the model were also significantly improved when normalization method was used at 450 to 1790 cm−1 sub-wavebands. For the calibration and cross-validation sample sets, the determination coefficients of the improved PLS-DA model were at 0.99 and 0.98, respectively. The mean square errors were 2.3 and 4%, resulting in a 90% accuracy of validation sample sets. This technology should provide fish markets an easy and reliable way to distinguish the adulterated and non-adulterated salmon.

Authentication and inference of seal stamps on Chinese traditional painting by using multivariate classification and near-infrared spectroscopy

Chinese traditional paintings occupy an important position in Chinese cultural heritage and it is very important for archeologist and artist to identify their authenticity, which is difficult to be realized. Near-infrared spectroscopy (NIRS) coupled with multivariate models was used for authenticating stamps of 12 seals on a Chinese traditional painting in this work. The robustness of linear and nonlinear multivariate models, i.e. partial least squares-discriminant analysis (PLS-DA) and support vector machine (SVM), were evaluated by adding 5 different levels of noise (from 1% to 5%) into 3 original NIR spectra of each the stamps. These spectral data with noise added were fused together with original spectra to establish identification models and then to evaluate the abilities of the two models to tolerate noise disturbance. Accuracies of 92.6% and 100% were yielded by linear PLS-DA and nonlinear SVM methods respectively. The results demonstrate the feasibility of multivariate approaches in authenticating stamps of seals on the Chinese traditional painting. It is also important and necessary to infer the approximate eras of seal stamps on Chinese traditional painting in archeological study. By comparing the Mahalanobis distances between the 12 stamps on the painting, hierarchical cluster analysis (HCA) was adopted to assist the inference of eras for those unknown seal stamps on the Chinese traditional painting. This work demonstrates that NIR spectroscopy combined with multivariate models can be utilized as a non-destructive approach for authentication of stamps on Chinese traditional painting. HCA can also provide useful information to speculate the time period of the stamps of unknown seals on the Chinese traditional painting.

Discrimination of Polygoni Multiflori radix and Cynanchi Auriculati radix using ultra-high performance liquid chromatography fingerprints and chemical pattern recognition.

In this work, a strategy was proposed to discriminate Polygoni Multiflori Radix (PMR) and its adulterant (Cynanchi Auriculati Radix, CAR). Ultra-high performance liquid chromatography (UHPLC) fingerprints were established to analyze samples containing PMR, CAR and mixtures simultaneously. Multivariate classification methods were applied to analyze the obtained UHPLC fingerprints, including principal component analysis (PCA), partial least square discriminant analysis (PLS-DA), soft independent modeling of class analogy (SIMCA), support vector machine discriminant analysis (SVMDA) and counter-propagation artificial neural network (CP-ANN). A plot of PCA score showed that PMR and CAR samples belonged to separate clusters (PMR class and CAR class), and samples of mixtures were located near PMR or CAR classes. Analysis by PLS-DA, SVMDA and CP-ANN performed well for recognition and prediction in terms of PMR and CAR samples. Moreover, the PLS-DA method performed best in the detection of adulterated samples, even if the adulterant was about 25%.