Qualitative analysis of green coffee by infrared spectrometry

Principal component analysis and linear discriminant analysis (LDA) were applied to the mid-infrared spectra for the qualitative analysis of the variety of green coffee (Arabica and Robusta). It is shown that the KBr pellet technique in combination with the LDA method can successfully be used for the identification of sample origin.     

Recognition for avian influenza virus proteins based on support vector machine and linear discriminant analysis

Total 200 properties related to structural characteristics were employed to represent structures of 400 HA coded proteins of influenza virus as training samples. Some recognition models for HA proteins of avian influenza virus (AIV) were developed using support vector machine (SVM) and linear discriminant analysis (LDA). The results obtained from LDA are as follows: the identification accuracy (Ria) for training samples is 99.8% and Ria by leave one out cross validation is 99.5%. Both Ria of 99.8% for training samples and Ria of 99.3% by leave one out cross validation are obtained using SVM model, respectively. External 200 HA proteins of influenza virus were used to validate the external predictive power of the resulting model. The external Ria for them is 95.5% by LDA and 96.5% by SVM, respectively, which shows that HA proteins of AIVs are preferably recognized by SVM and LDA, and the performances by SVM are superior to those by LDA.     

Ensemble of linear models for predicting drug properties

We propose a new classification method for the prediction of drug properties, called random feature subset boosting for linear discriminant analysis (LDA). The main novelty of this method is the ability to overcome the problems with constructing ensembles of linear discriminant models based on generalized eigenvectors of covariance matrices. Such linear models are popular in building classification-based structure-activity relationships. The introduction of ensembles of LDA models allows for an analysis of more complex problems than by using single LDA, for example, those involving multiple mechanisms of action. Using four data sets, we show experimentally that the method is competitive with other recently studied chemoinformatic methods, including support vector machines and models based on decision trees. We present an easy scheme for interpreting the model despite its apparent sophistication. We also outline theoretical evidence as to why, contrary to the conventional AdaBoost ensemble algorithm, this method is able to increase the accuracy of LDA models.     

Pattern recognition for the study of structure—activity relationships: Uses of the adaptive least-squares method and linear discriminant analysis

The adaptive least-squares method (ALS) and linear discriminant analysis (LDA) were applied to structure—activity correlation studies including the antitumor activity of mitomycin derivatives, the relative binding affinities of 100 steroids for five receptors, assignment of the pharmacological category of 80 diarylmethane-derived drugs, and discrimination of the adverse reaction of 98 miscellaneous drugs that may induce liver and/or blood diseases. Generally, more satisfactory results were obtained by the use of ALS than by LDA, both in recognition and in leave-one-out predictions. However, LDA was not always inferior to ALS in the applications, especially those related to classification of independent categories.     

交叉响应传感器鉴别β受体激动剂及其混合体系

开发了一种鉴别β受体激动剂的新型阵列传感器。该传感器由8种传感物质构成,使用96孔板酶标仪采集响应数据,结合主成分分析(PCA)、分层聚类分析(HCA)、判别分析(LDA)等模式识别方法进行数据处理,对5类β受体激动剂及其混合物进行检测。PCA结果表明,该传感器主要是基于空间结构以及氢键作用实现对β受体激动剂的识别;HCA结果显示,93个分析样本归类正确;LDA结果显示,该传感器对于β受体激动剂识别的准确率达98.9%。本方法在β受体激动剂的检测中有潜在应用价值。     

Rapid classification of White Oak (Quercus alba) and Northern Red Oak (Quercus rubra) by using pyrolysis direct analysis in real time (DART™) and time-of-flight mass spectrometry

Thirty-four samples of Red Oak (Quercus rubra) and fifty samples of White Oak (Quercus alba) were analyzed by pyrolytic direct analysis in real time (DART) ionization coupled with time-of-flight (TOF) mass spectrometry. Although significant differences were not observed in the positive-ion mass spectra, the negative-ion mass spectra showed clear differences. Principal component analysis (PCA) and linear discriminant analysis (LDA) were calculated for the relative abundances of 11 peaks in the negative-ion mass spectra including peaks tentatively assigned as representing deprotonated acetic, malic, gallic, dimethoxycinnamic, and ellagic acids. Leave one out cross validation (LOOCV) was 100% successful in classifying the samples for both PCA and LDA.     

Application of Multiple-Source Data Fusion for the Discrimination of Two Botanical Origins of Magnolia Officinalis Cortex Based on E-Nose Measurements,E-Tongue Measurements,and Chemical Analysis

Magnolia officinalis Rehd. et Wils. and Magnolia officinalis Rehd. et Wils. var. biloba Rehd. et Wils, as the legal botanical origins of Magnoliae Officinalis Cortex, are almost impossible to distinguish according to their appearance traits with respect to medicinal bark. The application of AFLP molecular markers for differentiating the two origins has not yet been successful. In this study, a combination of e-nose measurements, e-tongue measurements, and chemical analyses coupled with multiple-source data fusion was used to differentiate the two origins. Linear discriminant analysis (LDA) and quadratic discriminant analysis (QDA) were applied to compare the discrimination results. It was shown that the e-nose system presented a good discriminant ability with a low classification error for both LDA and QDA compared with e-tongue measurements and chemical analyses. In addition, the discriminating capacity of LDA for low-level fusion with original data, similar to a combined system, was superior or equal to that acquired individually with the three approaches. For mid-level fusion, the combination of different principals extracted by PCA and variables obtained on the basis of PLS-VIP exhibited an analogous discrimination ability for LDA (classification error 0.0%) and was significantly superior to QDA (classification error 1.67–3.33%). As a result, the combined e-nose, e-tongue, and chemical analysis approach proved to be a powerful tool for differentiating the two origins of Magnoliae Officinalis Cortex.     

Discrimination of Chinese traditional soy sauces based on their physico-chemical properties

This work aimed to classify the categories (produced by different processes) and brands (obtained from different geographical origins) of Chinese soy sauces. Nine variables of physico-chemical properties (density, pH, dry matter, ashes, electric conductivity, amino nitrogen, salt, viscosity and total acidity) of 53 soy sauce samples were measured. The measured data was submitted to such pattern recognition as cluster analysis (CA), principal component analysis (PCA), discrimination partial least squares (DPLS), linear discrimination analysis (LDA) and K-nearest neighbor (KNN) to evaluate the data patterns and the possibility of differentiating Chinese soy sauces between different categories and brands. Two clusters corresponding to the two categories were obtained, and each cluster was divided into three subsets corresponding to three brands by the CA method. The variables for LDA and KNN were selected by the Fisher F-ratio approach. The prediction ability of all classifiers was evaluated by cross-validation. For the three supervised discrimination analyses, LDA and KNN gave 100% predications according to the sample category and brand.     

Forensic classification of ballpoint pen inks using high performance liquid chromatography and infrared spectroscopy with principal components analysis and linear discriminant analysis

Several varieties of blue ballpoint pen inks were analyzed by high performance liquid chromatography (HPLC) and infrared spectroscopy (IR). The chromatographic data extracted at four wavelengths (254, 279, 370 and 400 nm) was analyzed individually and at a combination of these wavelengths by the soft independent modeling of class analogies (SIMCA) technique using principal components analysis (PCA) to estimate the separation between the pen samples. Linear discriminant analysis (LDA) measured the probability with which an observation could be assigned to a pen class. The best resolution was obtained by HPLC using data from all four wavelengths together, differentiating 96.4% pen pairs successfully using PCA and 97.9% pen samples by LDA. PCA separated 60.7% of the pen pairs and LDA provided a correct classification of 62.5% of the pens analyzed by IR. The results of this study indicate that HPLC coupled with chemometrics provided a better discrimination of ballpoint pen inks compared to IR. The need to develop a suitable IR method for analysing blue ballpoint pen inks has been emphasized and it is hoped that the development of such a method would indeed provide a valuable tool for the non-destructive analysis of blue ballpoint pen ink samples for forensic purposes.     

Classification of Aurantii Fructus samples by multivariate analysis

Twenty commercial samples of Aurantii Fructus Immaturus (Poncitrus trifoliata) and 30 of Aurantii Fructus Maturus (Citrus aurantium and C. wilsonii) were collected from the Taiwan and China herbal markets. The contents of 12 constituents in these samples were determined by HPLC and were used to assess the potential relationships with their plant origins. Multivariate analysis including principal component analysis (PCA), cluster analysis (CA), and linear discriminant analysis (LDA) were used as classification procedures. Natural groupings of the samples divided into three sets successfully, 20 P. trifoliata, 15 C. aurantium, and 15 C. wilsonii, were observed by using PCA and CA. The application of LDA gave correct assignation percentages of 100.0% for all three groups.     

Varietal discrimination of Australian wines by means of mid-infrared spectroscopy and multivariate analysis

This study outlines the use of mid-infrared (MIR) spectroscopy combined with principal component analysis (PCA) and linear discriminant analysis (LDA) for the varietal classification of commercial red and white table wines. Three red varieties (Cabernet Sauvignon, Shiraz and Merlot) and four white varieties (Chardonnay, Riesling, Sauvignon Blanc and Viognier) were sourced from different wine regions in Australia. Wine samples were scanned in transmission on a FOSS WineScan FT 120 from wave numbers 926 to 5012 cm⁻¹. All samples were sourced from the 2006 vintage and had not been blended with any other variety or wine from other regions. Spectral data were reduced to a small number of principal components (PCs) and LDA was then performed to successfully separate the wines into the different varieties. To test the robustness of the LDA models developed for the red wines, a set of red wines scanned in 2005 were used. Correct classification of over 95% was achieved for the validation set.     

A pattern recognition based fluorescence quenching assay for the detection and identification of nitrated explosive analytes

Herein we report a differential array of micelle-solubilized fluorophores for the detection and identification of small nitrated analytes, such as the explosives TNT, tetryl, RDX and HMX. The quenching ability of the analytes can be used to correlate their analyte identity, wherein the quenching patterns generated from the differential array are used in linear discriminant analysis (LDA). LDA results in a well-clustered two-dimensional plot, and a jack-knife analysis of the data suggests that this system can be used to identify unknown samples of analyte with 96 % accuracy and with a detection limit of 19 muM.     

Comparison of methods for outlier detection and their effects on the classification results for a particular data base

One of the drawbacks for using linear discriminant analysis (LDA) is the presence of outliers. Some methods of detecting outliers are compared and applied to a particular data base. When multivariate methods (multinormal distribution procedure and Hawkins' procedure) were applied, the two subsets produced did not differ greatly. Assumptions needed for the application of LDA were evaluated for each subset. Classification ability, feature selection and prediction ability were considered for each subset. Results for each subset were quite different. Hawkins' procedure seems the better method for detecting outliers.     

Meat species identification by linear discriminant analysis of capillary electrophoresis protein profiles

The objective of this study was to utilize linear discriminant analysis (LDA) in the interpretation of capillary electrophoresis-sodium dodecyl sulfate polymer-filled capillary gel electrophoresis (CE-SDS) meat protein profiles for the identification of meat species. The specific objectives were 1) to collect quantitative data on water-soluble and saline-soluble proteins of different meat species obtained by CE-SDS and 2) to apply LDA on collected CE-SDS protein data for the development of a pattern recognition statistical model useful in the differentiation of meat species. Samples were raw beef top and eye round, boneless fresh pork ham and loin, turkey leg and breast meat, and mechanically deboned turkey meat collected on six different occasions, making a total of 42 samples. Additionally, 14 samples were used as test samples to determine the classification ability of the procedure. Quantitative protein data obtained by CE-SDS was used to generate separate LDA models for either water- or saline-soluble protein extracts. Although a saline solution was a more efficient meat protein-extracting agent, as shown by a higher total protein concentration and a larger number of peaks, water-soluble CE-SDS protein profiles gave more distinctive discrimination among meat species. The correct classification given by LDA on water-soluble protein data was 100% for all meat species, except pork (94%). Conversely, the correct classification on saline-soluble protein data was 88% for beef and mechanically deboned turkey meat, and 94% and 100% for turkey and pork meat, respectively. LDA proved to be a useful pattern recognition procedure in the interpretation of CE-SDS protein profiles for the identification of meat species.     

Water's polyamorphic transitions and amorphization of ice under pressure

Transformations of water's high density amorph (HDA) to low density amorph (LDA) and of LDA's to cubic ice (Ic) have been studied by in situ thermal conductivity kappa measurements at high pressures. The HDA to LDA transformation is unobservable at p of 0.07 GPa, indicating that, for a fixed heating rate, an increase in pressure increases the temperature of HDA to LDA transformation and decreases that of LDA to ice Ic, causing thereby the two transformations to merge, and HDA appears to convert directly to ice Ic. Thus either LDA forms but converts extremely rapidly to ice Ic, or LDA does not form. At a fixed p and T, in the range of pressure amorphization of hexagonal ice, kappa continues to decrease with time. Therefore, the amorphization of ice Ih is kinetically controlled. When HDA at 1 GPa was heated from 130 to 157 K and densified to very HDA, its kappa increased by 3%. Our findings and a scrutiny of earlier reports show that a reversible transition between HDA and LDA does not occur at approximately 135 K and approximately 0.2 GPa. Since there is no unique HDA, it is difficult to justify the conjecture for a second critical point for water.     

Enhancing Classification Performance of Multisensory Data through Extraction and Selection of Features

Linear discriminant analysis (LDA) has been widely used in the classification of multi sensor data fusion. This paper discusses the performance of LDA when the classifications were performed based on feature extraction and feature selection methods. Comparisons were also made based on single sensor modality. These strategies were studied using a honey dataset along with two types of sugar concentration collected from two types of sensors namely electronic nose (e-nose) and electronic tongue (e-tongue). Assessment of error rate was achieved using the leave-one-out procedure.