基于电子鼻的不同去势猪肉风味品质评价

实验分别对免疫去势公猪肉、手术去势公猪肉和完全公猪肉进行电子鼻检测,并采用主成分分析、线性判别式分析和交互验证判别分析分别对电子鼻15s、30s和60s响应值进行统计处理。结果表明,主成分分析效果不好,三个处理组几乎完全重叠;线性判别式分析结果显示,采用15s响应值其区分效果及聚类效果最好,完全公猪组的气味显著地区别于免疫去势和手术去势组,且免疫去势组的气味与手术去势组相似;对15s、30s和60s响应值进行交互验证判别分析,总体正确率依次为90.0%、83.3%、66.7%,由各组的正确率可知,完全公猪组的正确率最高,正确率稍低的30s和60s响应值的分析结果显示,手术去势组和免疫去势组较易混淆,说明这两组气味相似。综上所述,电子鼻的检测结果显示,手术去势组和免疫去势组的气味相似,且均与完全公猪组有较大差异。     

Application of discriminant analysis to parameters describing the drape for two types of woven fabrics

A total of 33 commercial drapery and lining fabrics were used to determine the drape indicators drape ratio and R-factor. The slope of a plot of one indicator against the other was found not to afford complete characterization of drape shape for fabrics of different commercial use. In this work, further six parameters describing the drape, proposed by the authors to discriminate drape shapes were also calculated. Discriminant analysis of the data revealed that a linear combination of various parameters allowed two types of woven fabrics (viz. drapery and linen) to be distinguished. The discriminant function used accurately classified 75.76% of the fabrics studied.     

Can we identify key characteristics associated with grazing-management dairy systems from survey data?

Discriminant analysis was used to identify farms using confinement and grazing-production systems from mail survey data of 2074 dairy farmers in Pennsylvania, Vermont, Virginia, and North Carolina. Survey respondents included 45.1% of the farms using confinement management; 13.5% of farms practicing intensive grazing, defined as moving cows to new pasture at least every 3 d; and 41.4% of farms using nonintensive grazing. Farmers using confinement management had significantly more cows, higher milk production, more crop acreage, higher debt, used automatic takeoff milking units (ATO), fed total mixed rations (TMR), and were more satisfied. In general, dairy farmers who grazed their milking cows had smaller herds, fewer acres, but had more acres per cow and made less use of technology. However, farmers practicing intensive grazing were significantly younger, more educated, less experienced, more likely to use computers, and farmed less acreage than other graziers or farmers on confinement farms. The discriminant function correctly classified 70% of the total sample when divided into confinement and overall grazing categories. However, the discriminant function correctly classified only 36% of intensive-grazing farms in comparison to confinement farms. Significant variables identified using ordinary least squares as being related to confinement management were milk per cow, acres of corn, use of ATO and TMR, debt greater than 40%, and residence in North Carolina. Significant variables associated with grazing management were acres of pasture, future use of pasture, education, and residence in Vermont. The analysis indicated that the discriminant function could correctly classify confinement and nonintensive-grazing management but was unable to reliably differentiate between confinement and intensive-grazing farms.     

腐乳理化特性分析和模式识别研究

分析了27种腐乳的理化特性，并以此为依据，进行模式识别研究，结果将27种腐乳聚成了四川腐乳和其他腐乳2类，在6个理化指标的基础上建立了3个主成分模型，并得到了腐乳的判别方利D=-0．080x1＋1．746x2＋0．396x3＋0．348x4＋0．140x5-0．322x6-0．580，总判别止确率为96．3％，为控制腐乳品质提供了一定的理论依据。     

Application of a statistical approach to the classification of honey by geographic origin

Forty eight honeys from the La Rioja region of Spain were analysed to establish their quality. Fourteen legally required parameters of quality control were measured. Samples were obtained from two geographic areas of production: Valley and Sierra. These areas present different agroclimatic conditions and, thus, different flora. Classification of these honeys according to their geographic origin was achieved by applying multivariate statistical analysis to the chemical and physical data. Acidity (free acidity and pH), mineral content (electrical conductivity and ash) and factors related to degree of freshness (hydroxymethylfurfural and diastatic activity), were found to be the most important features for classification. Determination of the geographic origin of a La Rioja honey (Valley or Sierra) was possible with 83% accuracy using only these legally required quality control parameters; pollen studies were not necessary to achieve this objective.     

基于HPLC指纹图谱鉴别四个不同品种的金银花叶

利用金银花叶的化学物质群差异,开发金银花不同品种之间的鉴别技术。以灰毡毛忍冬、红腺忍冬、“北花1号”金银花、四季金银花的叶片为实验材料,采用高效液相色谱仪(HPLC)构建金银花叶的化学指纹图谱,从中提取了34个特征色谱峰,结合主成分分析、系统聚类分析等多变量统计分析方法,比较分析了4种不同品种的金银花叶的化学物质群差异。结果显示,前两个主成分累计表征了51.3%的原始变量,在主成分得分图上4个不同品种的金银花样品呈现各自相对独立的空间分布特征。聚类分析可将32个供试样品按品种来源及其相似程度聚为4类。相较于四季金银花,“北花1号”品种与之具有最相似化学特征,其次依次为红腺忍冬和灰毡毛忍冬。指纹图谱中的第5、6、7(绿原酸)、8、11、14、16、18、24、29号等10个色谱峰,是各品种金银花叶化学差异的主要来源,可以进一步开发成为稳定的化学标记。本研究所建立的HPLC指纹图谱及其分析方法,可以用于灰毡毛忍冬、红腺忍冬、“北花1号”金银花、四季金银花等不同品种金银花之间的鉴别分析。     

Geographical characterisation of honeys according to their mineral content and antioxidant activity using a chemometric approach

In this article, discrimination models are presented, relating the origin of honey samples to several variables, being the concentrations of different cations and anions in the honey samples measured by ion chromatography, and parameters that measure/reflect the antioxidant activity of the honey samples. The unsupervised method, principal component analysis, and supervised discrimination methods, such as linear and quadratic discriminant analysis, and classification and regression trees (CART), were applied to evaluate the existence of data patterns and the relationship between geographical origin and the measured parameters. The model with the best predictive ability (%CCR_TEST = 66.67%), the best overall % specificity (80%) and the best overall % sensitivity (67%) was found to be CART. It was proven that the mineral content and parameters analysed can provide enough information for the geographical characterisation and discrimination of honey.     

Authentication of Coffee Bean Variety by Near-infrared Reflectance Spectroscopy of Dried Extract

The potential of near-infrared (NIR) reflectance spectroscopy for discriminating between coffee beverage prepared from pure Arabica, pure Robusta and blends of these two varieties was investigated. Dried beverages were produced by both lyophilisation and air-drying under vacuum on glass-fibre filter paper. Spectral collections were treated by principal component and factorial discriminant analyses. Using the wavelength range 1100–2498 nm, only three of 65 test samples were misclassified when the filter paper approach was used. When freeze-dried coffee beverages were analysed, nine of the 65 test samples were misclassified. The basis for this discrimination appears to involve caffeine and/or other alkaloids.     

不同品种红茶及茶膏的Fisher判别分析

为探究Fisher判别分析法对不同品种红茶及茶膏种类品质区分的可行性,针对不同品种红茶及滇红碎茶茶膏,以蛋白质、总糖、茶红素、茶黄素、茶褐素及9 种矿物质元素为变量进行判别分析。结果表明：采用Fisher判别分析法评定不同品种红茶及不同加工工艺滇红碎茶茶膏的真实归属及品质是可靠的。红茶茶叶及茶膏品质评定中采用Fisher判别分析有利于降低传统感官评定过程中人为主观因素的干扰,避免模糊性和不确定性,保证评审的稳定和可靠性。     

LDA优化电子鼻传感器阵列的研究

利用PEN3电子鼻系统对6个糖酸比不同的乳饮料样品进行检测,采用线性判别分析(LDA)对传感器响应值进行分析,确定优化传感器阵列方法,并将各优化结果进行对比,最终确定阵列优化结果,使电子鼻可以用更少的传感器达到更好的分类效果,为电子鼻传感器阵列优化提供了新的思路和方法。     

基于主成分分析和判别分析的稻米产地溯源

以提高矿物元素对大米产地溯源的稳定性和准确性,找寻表征小范围相似地域特性的溯源指标为目的。采用原子吸收光谱法（AAS）分析吉林省松原市三大主产区10个产地100个大米样品中的矿物元素含量,对所得矿物元素含量数据进行差异分析、判别分析、主成分分析和聚类分析。试验结果为实现了松原市三大主产区大米产地溯源,正确判别率为100%。松原市矿物元素溯源指标的筛选主成分分析结果：第一主成分主要由Zn、K、Mg、Na、Ca和Mn 等6种元素构成,贡献率最大,占47.176%。判别分析验证主成分分析和聚类分析的准确性,其正确判别率为100%。利用矿物元素实现小范围产地溯源,并获得吉林省松原市溯源指标：Zn、K、Mg、Na、Ca和Mn 6种元素。     

Characterizing ruminal acidosis risk: A multiherd,multicountry study

A multicenter observational study was conducted on early lactation Holstein cows (n = 261) from 32 herds from 3 regions (Australia, AU; California, CA; and Canada, CAN) to characterize their risk of acidosis into 3 groups (high, medium, or low) using a discriminant analysis model previously developed. Diets ranged from pasture supplemented with concentrates to total mixed ration (nonfiber carbohydrates = 17 to 47 and neutral detergent fiber = 27 to 58% of dry matter). Rumen fluid samples were collected <3 h after feeding and analyzed for pH, and ammonia, d- and l-lactate, and volatile fatty acid (VFA) concentrations. Eigenvectors were produced using cluster and discriminant analysis from a combination of rumen pH, and ammonia, d-lactate, and individual VFA concentrations and were used to calculate the probability of the risk of ruminal acidosis based on proximity to the centroid of 3 clusters. Bacterial 16S ribosomal DNA sequence data were analyzed to characterize bacteria. Individual cow milk volume, fat, protein, and somatic cell count values were obtained from the closest herd test to the rumen sampling date (median = 1 d before rumen sampling). Mixed model analyses were performed on the markers of rumen fermentation, production characteristics, and the probability of acidosis. A total of 26.1% of the cows were classified as high risk for acidosis, 26.8% as medium risk, and 47.1% as low risk. Acidosis risk differed among regions with AU (37.2%) and CA (39.2%) having similar prevalence of high-risk cows and CAN only 5.2%. The high-risk group had rumen phyla, fermentation, and production characteristics consistent with a model of acidosis that reflected a rapid rate of carbohydrate fermentation. Namely, acetate to propionate ratio (1.98 ± 0.11), concentrations of valerate (2.93 ± 0.14 mM), milk fat to protein ratio (1.11 ± 0.047), and a positive association with abundance of phylum Firmicutes. The medium-risk group contains cows that may be inappetant or that had not eaten recently or were in recovery from acidosis. The low-risk group may represent cattle that are well fed with a stable rumen and a slower rumen fermentation of carbohydrates. The high risk for acidosis group had lower diversity of bacteria than the other groups, whereas CAN had a greater diversity than AU and CA. Rumen fermentation profile, abundance of ruminal bacterial phyla, and production characteristics of early lactation dairy cattle from 3 regions were successfully categorized in 3 different acidosis risk states, with characteristics differing between acidosis risk groups. The prevalence of acidosis risk also differed between regions.     

Classifying the fertility of dairy cows using milk mid-infrared spectroscopy

The objective of this study was to investigate the potential of milk mid-infrared (MIR) spectroscopy, MIR-derived traits including milk composition, milk fatty acids, and blood metabolic profiles (fatty acids, β-hydroxybutyrate, and urea), and other on-farm data for discriminating cows of good versus poor likelihood of conception to first insemination (i.e., pregnant vs. open). A total of 6,488 spectral and milk production records of 2,987 cows from 19 commercial dairy herds across 3 Australian states were used. Seven models, comprising different explanatory variables, were examined. Model 1 included milk production; concentrations of fat, protein, and lactose; somatic cell count; age at calving; days in milk at herd test; and days from calving to insemination. Model 2 included, in addition to the variables in model 1, milk fatty acids and blood metabolic profiles. The MIR spectrum collected before first insemination was added to model 2 to form model 3. Fat, protein, and lactose percentages, milk fatty acids, and blood metabolic profiles were removed from model 3 to create model 4. Model 5 and model 6 comprised model 4 and either fertility genomic estimated breeding value or principal components obtained from a genomic relationship matrix derived using animal genotypes, respectively. In model 7, all previously described sources of information, but not MIR-derived traits, were used. The models were developed using partial least squares discriminant analysis. The performance of each model was evaluated in 2 ways: 10-fold random cross-validation and herd-by-herd external validation. The accuracy measures were sensitivity (i.e., the proportion of pregnant cows that were correctly classified), specificity (i.e., the proportion of open cows that were correctly classified), and area under the curve (AUC) for the receiver operating curve. The results showed that in all models, prediction accuracy obtained through 10-fold random cross-validation was higher than that of herd-by-herd external validation, with the difference in AUC ranging between 0.01 and 0.09. In the herd-by-herd external validation, using basic on-farm information (model 1) was not sufficient to classify good- and poor-fertility cows; the sensitivity, specificity, and AUC were around 0.66. Compared with model 1, adding milk fatty acids and blood metabolic profiles (model 2) increased the sensitivity, specificity, and AUC by 0.01, 0.02, and 0.02 unit, respectively (i.e., 0.65, 0.63, and 0.678). Incorporating MIR spectra into model 2 resulted in sensitivity, specificity, and AUC values of 0.73, 0.63, and 0.72, respectively (model 3). The comparable prediction accuracies observed for models 3 and 4 mean that useful information from MIR-derived traits is already included in the spectra. Adding the fertility genomic estimated breeding value and animal genotypes (model 7) produced the highest prediction accuracy, with sensitivity, specificity, and AUC values of 0.75, 0.66, and 0.75, respectively. However, removing either the fertility estimated breeding value or animal genotype from model 7 resulted in a reduction of the prediction accuracy of only 0.01 and 0.02, respectively. In conclusion, this study indicates that MIR and other on-farm data could be used to classify cows of good and poor likelihood of conception with promising accuracy.     

Pombe: A gene-finding and exon-intron structure prediction system for fission yeast

A special program developed by the authors, called Pombe, identifies protein coding regions in the Schizosaccharomyces pombe genome. Linear discriminant analysis was applied to predict 5′-terminal, internal, 3′-terminal exons (coding-exon) and introns. The accuracy of the prediction was tested by cross verifications. The sensitivity, specificity and correlation coefficient for the internal exon prediction were 98·5%, 99·9% and 98·3% respectively at the nucleotide level. Open reading frames were studied and used to predict intron-less genes: 99·0% of such genes were identified with correct stopping sites. The gene structure was determined by dynamic programming and the prediction achieved 97·0% correlation coefficient at the nucleotide level. The program is available at http://clio.cshl.org/genefinder. © 1998 John Wiley & Sons, Ltd.     

基于主成分分析和聚类分析的海沃德猕猴桃品质指标综合评价

为更准确评价猕猴桃品质,选取陕西省眉县的157 个海沃德猕猴桃,对果实的单果质量、长轴、短轴、厚度、体积、果皮颜色和糖度、酸度、硬度9 个分级指标进行了描述统计和相关分析,采用主成分分析法建立综合得分数学模型,对综合得分进一步做K-means聚类分析,最后利用Fisher判别分析法对样品重新进行聚类以验证K-means聚类分析方法的可靠性。结果表明,除体积与单果质量间差异不明显外,其余各分级指标之间均存在显著差异;按综合得分将样品聚为3 类：优为0.10～1.39,中为－0.44～0.09,差为－1.27～－0.46;判别分析对聚类结果的正确率达到98.72%,所以两者具有较高的一致性。     

主成分和判别分析在清香型白酒产地溯源中的应用

采用气相色谱法分析不同产地清香型白酒中乙酸乙酯、乳酸乙酯、乙酸等12种主体香味成分的含量,利用SPSS22.0软件对数据进行主成分和判别分析,可以将12个指标提取3个主成分,建立3个判别典型函数。利用建立判别函数可以对5个不同产地的78个白酒样品100%进行正确判别。该方法为不同产地白酒的溯源和真伪鉴别提供一条新的途径。     

电子鼻快速检测区分羊肉中的掺杂鸡肉

为实现掺假羊肉的快速、客观检测,利用电子鼻定性和定量分析混入鸡肉的掺假羊肉糜。单因素试验表明顶空体积、载气流速、样品量和顶空生成时间对电子鼻传感器的响应影响极显著;主成分分析确定了电子鼻检测的较佳条件：样品量10 g、载气流速200 mL/min、顶空容积250 mL及顶空生成时间30 min。在此条件下检测混入鸡肉的掺假羊肉,结果发现采用主成分分析时,掺入鸡肉的比例随主成分一降低而增大,但相邻比例彼此重叠,难以有效区分;采用典则判别分析时,混入不同比例鸡肉的羊肉糜样品能较好地区分;采用主成分回归分析和偏最小二乘回归分析建立的定量预测模型（R2>0.95）能有效预测混入的鸡肉比例。电子鼻在混入鸡肉的掺假羊肉鉴别中具有可行性,论文可为羊肉掺假鉴别提供理论依据。     

基于指纹图谱和化学计量学的浓香型白酒分类研究

该研究以市售畅销的7个品牌、不同等级的浓香型白酒为研究对象,利用气相色谱-质谱联用（GC-MS）技术建立了浓香型白酒中风味成分的指纹图谱,结合相似度分析、主成分分析（PCA）和判别分析（DA）对不同品牌的浓香型白酒样品进行了有效区分和鉴别。结果表明,不同品牌间样品的相似度存在一定差异;PCA表明前三个主成分累计方差贡献率达到89.71%,能对样品进行聚类和区分;利用判别分析可以将不同产地的酒样区分开,正确率为100%。综上,相同品牌不同等级的白酒具有明显的相关性,不同品牌白酒可以利用指纹图谱结合化学计量学方法进行鉴别和分类,为白酒质量控制及真伪鉴定提供了一种新方法。     

因子、聚类及判别分析在烟叶风格特色评价中的应用

为合理评价烟叶风格特征及不同风格特征烟叶区域分类,采用因子、聚类及判别分析相结合的方法,对河南31个产烟县的169个烟叶样品的风格特征指标进行了分析。结果表明,河南烟叶浓香型风格为尚显著;香韵以干草香、焦甜香与焦香为主;河南烟叶的香型与香气状态、香韵中的焦甜香、焦香、烟气浓度呈极显著正相关,与劲头呈显著正相关;采用因子分析方法提取了5个主因子,根据因子总得分进行聚类分析,将31个产烟县分为三类。建立Fisher判别函数评价模型对聚类结果进行判定,判别结果和聚类分析一致性为100.0%。证明这3种方法相结合用于烟叶风格特色分类评价可行性强,效果较好。