This study was carried out with the purpose of discovering what factors are predictive of money management behavior; Plan, Implement, and Evaluative Feedback. The data used in this study was subset data collected during 1988, under the NC-167 project entitled "Family Resource Utilization as a Factor in Determining Economic Well Being of Rural Families". Three hundred and seven financial managers in families from Arizona completed and returned the questionnaire used in this study. It was found that (1) the power money attitude, the inadequacy money attitude, and gender were predictor variables of plan behavior; (2) the inadequacy money attitude and age were predictor variables of evaluative feedback.

Clustered failure time data often arise in biomedical and clinical studies where potential correlation among survival times is induced in a cluster. In this thesis, we develop a class of marginal models for right censored clustered failure time data and propose a novel generalized estimating equation approach in a likelihood-based context. We first investigate a semiparametric proportional hazards model for clustered survival data and derive the large sample properties of the regression estimators. The finite sample studies demonstrate that the good applicability of the proposed method as well as the substantial efficiency improvement in comparison with the existing marginal model for clustered survival data. Another important feature of failure time data we will consider in this thesis is a possible fraction of cured subjects. To accommodate the potential cure fraction, we consider a proportional hazards mixture cure model for clustered survival data with long-term survivors and develop a set of estimating equations by incorporating working correlation matrices in an EM algorithm. The dependence among the cure statuses and among the survival times of uncured patients within clusters are modeled by working correlation matrices in the estimating equations. For the parametric proportional hazards mixture cure model, we show that the estimators of the regression parameters and the parameter in the baseline hazard function are consistent and asymptotically normal with a sandwich covariance matrix that can be consistently estimated. A numerical study presents that the proposed estimation method is comparable with the existing parametric marginal method. We also extend the proposed generalized estimating equation approach to a semiparametric proportional hazards mixture cure model where the baseline survival function is nonparametrically specified. A bootstrap method is used to obtain the variances of the estimates. The proposed method is evaluated by a simulation study from which we observe a noticeable efficiency gain of the proposed method over the existing semiparametric marginal method for clustered failure time data with a cure fraction. / Thesis (Ph.D, Mathematics & Statistics) -- Queen's University, 2013-01-30 21:23:48.968

The objective of this thesis is to investigate various factors that influence the job performance and promotion of DOD civilian workers. The data used in this study were drawn from the Department of Defense Civilian Personnel Data Files provided by the Defense Manpower Data Center (DMDC). The initial data was restricted to employees who were initially hired in 1995 and stayed in service until 2003 and were paid under the General Schedule (GS) pay system. Three general performance measures were used: compensation (salary), annual performance ratings and promotions. Multivariate models were specified and estimated for each of these performance measures. The results indicate that females receive lower annual and hourly compensation and are less likely to be promoted than men even though they receive better performance ratings. The results also indicate that minorities are paid less and are less likely to be promoted than majority workers while veterans are paid more, perform better, and are more likely to become supervisors. The models also reveal that performance rating is a weak measure of productivity and that more highly educated employees are paid more and more likely to be promoted more even if they are not always the best performers.

Student no:7719196 PhD thesis 2005 Faculty of Science, School of Animal, Plant and Environmental Sciences. / This study aimed to investigate the phylogeny of Cineraria L. to elucidate its position in the tribe Senecioneae (Asteraceae), to delimit its generic and species boundaries, explore species relationships (infrageneric structure) and produce a monograph of the genus. It also aimed to examine its distribution, phytogeographic affinities, levels of rarity and endemism in Cineraria, to identify some of the factors contributing to rarity in Cineraria and to highlight implications for conservation. Phylogenetic analyses were performed using morphological and molecular (DNA sequence) characters to elucidate relationships within the genus and between Cineraria and selected related genera in the subtribe Senecioninae. The phylogenetic species concept was applied - suites of diagnostic characters were used to characterise species. The phenetic approach, using Cluster Analysis and Principal Coordinates Analysis, was applied to investigate variation in two highly variable species, C. deltoidea Sond. and C. lobata L’Hér. Species distributions were mapped and the number of species per degree square was plotted for southern Africa to identify centres of diversity and endemism. Rare species were identified and categorised according to Rabinowitz’s criteria of geographic range, habitat specificity and local population size. Cineraria now has a more homogenous generic concept, characterised as herbs or subshrubs with palmately veined leaves, radiate, calyculate capitula, penicillate style apices and obovate, compressed cypselae with two distinct margins or wings and a substantial carpopodium. Eleven species have been removed from the genus and two new genera, Bolandia Cron and Oresbia Cron & B.Nord. have been established to accommodate three of the species. Two species have been reassigned to and another reinstated in Senecio L. The affinities of five species remain unresolved. Cineraria now comprises 35 species with four new subspecies and two new varieties recognised. Eight species have been placed in synonymy and five new species have been described during this revision. The status of C. deltoidea as a single, highly variable species, widespread throughout the eastern mountains of Africa, has been confirmed. Cineraria lobata has been shown to be a very variable species and the geographic and morphological variation has been formally (and informally) recognised. No infrageneric classification has been applied to Cineraria as a robust phylogeny of all the species has yet to be hypothesised. A southern African origin for Cineraria in the Western and/or Eastern Cape is postulated, based on the current distribution of the sister genus Bolandia and Cineraria mollis DC. iv Cineraria appears to have undergone rapid speciation fairly recently, as indicated by the lack of variation in the molecular data analysed, with reticulate evolution playing an important role in its evolutionary history (as seen by the lack of congruence between the chloroplast and nuclear DNA sequences). The centre of diversity of Cineraria is the KwaZulu-Natal Midlands, part of the Maputaland-Pondoland Centre of endemism in southern Africa. Cineraria has an afromontane affinity, and fifteen species endemic to specific mountains or regions of endemism and five near-endemics have been identified in Cineraria. Eleven species have been shown to be rare (in the sense of low abundance, restricted range and high habitat specificity), however only five are considered to be threatened as indicated by IUCN Red Data Criteria. However, at least seven species are Data Deficient and require further investigation. Causes of rarity in Cineraria are linked to narrow habitat specificity, particularly soil or rock type and/or altitudinal range.

The horseshoe prior has been shown to successfully handle high-dimensional sparse estimation problems. It both adapts to sparsity efficiently and provides nearly unbiased estimates for large signals. In addition, efficient sampling algorithms have been developed and successively applied to a vast array of high-dimensional sparse estimation problems. In this dissertation, we investigate the prediction performance of the horseshoe prior in sparse regression, and extend the horseshoe prior to two multivariate settings.<br><br>We begin with a study of the finite sample prediction performance of shrinkage regression methods, where the risk can be unbiasedly estimated using Stein's approach. We show that the horseshoe prior achieves an improved prediction risk over global shrinkage rules, by using a component-specific local shrinkage term that is learned from the data under a heavy-tailed prior, in combination with a global term providing shrinkage towards zero. We demonstrate improved prediction performance in a simulation study and in a pharmacogenomics data set, confirming our theoretical findings.<br><br>We then shift to extending the horseshoe prior to handle two high-dimensional multivariate problems. First, we develop a new estimator of the inverse covariance matrix for high-dimensional multivariate normal data. The proposed graphical horseshoe estimator has attractive properties compared to other popular estimators. The most prominent benefit is that when the true inverse covariance matrix is sparse, the graphical horseshoe estimator provides estimates with small information divergence from the sampling model. The posterior mean under the graphical horseshoe prior can also be almost unbiased under certain conditions. In addition to these theoretical results, we provide a full Gibbs sampler for implementation. The graphical horseshoe estimator compares favorably to existing techniques in simulations and in a human gene network data analysis.<br><br>In our second setting, we apply the horseshoe prior to the joint estimation of regression coefficients and the inverse covariance matrix in normal models. The computational challenge in this problem is due to the dimensionality of the parameter space that routinely exceeds the sample size. We show that the advantages of the horseshoe prior in estimating a mean vector, or an inverse covariance matrix, separately are also present when addressing both simultaneously. We propose a full Bayesian treatment, with a sampling algorithm that is linear in the number of predictors. Extensive performance comparisons are provided with both frequentist and Bayesian alternatives, and both estimation and prediction performances are verified on a genomic data set.

O sistema orçamentário é um artefato contábil relevante para a tomada de decisão diante de um contexto de alta complexidade apresentado na realidade contemporânea, pois representa uma antecipação da visão da empresa, traduzindo sua estratégia e oferecendo diretrizes de conduta. Sua composição e as razões para o seu uso (como forma de comunicação das metas, avaliando desempenho, sendo um instrumento de planejamento operacional e de formação da estratégia) parecem estar relacionadas às variáveis contingenciais que afetam a empresa, tais como a incerteza ambiental percebida, a estratégia em foco, a estrutura organizacional, a tecnologia empregada em seu sistema de informação e o seu porte. O presente estudo, portanto, possui como finalidade compreender os atributos do sistema orçamentário a partir das variáveis contingenciais, considerando determinados níveis de desempenho. Metodologicamente, trata-se de um estudo descritivo e causal, utilizando-se de coleta de dados com questionário previamente formulado, aplicado para os controllers ou responsáveis pela controladoria nas indústrias paranaenses cadastradas em 2006 na FIEP- Federação das Indústrias do Estado do Paraná. Adicionalmente, refere-se a um estudo ex post facto, em condições de campo, transversal e utilizando-se de técnicas estatísticas de análise multivariada, mais especificamente testes não-paramétricos, correlação de Pearson, análise fatorial, modelagem de equações estruturais e clusters. Por subsidiar sua investigação com a abordagem contingencial, o presente estudo pode ser caracterizado como sendo uma pesquisa de sistemas, relacionando as diversas variáveis em análise com o desempenho percebido para a identificação de arranjos contingenciais. Os resultados apresentados apontam para a aceitabilidade das cinco hipóteses apresentadas na tese, ratificando a influência do ambiente nos fatores contingenciais intra-organizacionais, a relação dos fatores contingenciais e o sistema orçamentário da empresa, a influência das variáveis contingenciais no desempenho, a associação existente entre os atributos orçamentários e o desempenho e, por fim, a presença de arranjos (fits) entre as variáveis contingenciais e os atributos do sistema orçamentário, em virtude da apresentação de um desempenho superior. A presente pesquisa empírica revela que independente do porte, indústrias paranaenses que possuem um maior grau de aderência ao sistema orçamentário, conscientes das razões para o seu uso e que apresentam tarefa rotineira em seus processos exibiram alto desempenho organizacional e orçamentário. Logo, conclui-se que sob certas condições há indícios que o orçamento é benéfico à empresa, revelando uma relação custo-benefício positiva. / The budgetary system is a relevant accounting tool to the decision-making within the high complex context of the contemporary reality, since it represents a prospect of the business vision, translating its strategy and offering conduct guidance. The budgetary system\'s composition and the reasons for its use (as a form of targets communication, performance evaluation and being an instrument for operational planning and strategic formation) seem to be related to contingent variables that affect a business, such as perceived environmental uncertainty, focus strategy, the organizational structure, the information system technology employed and its size. The present study, therefore, intends to understand the budgetary system attributes considering contingent variables and certain performance levels. Methodologically, it is a descriptive and causal study, witch uses data from previously formulated surveys, aimed to the controllers or to the responsible for the control department of the Paraná industries registered at FIEP - Federação das Indústrias do Estado do Paraná in 2006. Additionally, it is an ex post facto study, in field conditions, transversal, and using multivariate analysis statistical techniques, more specifically non parametric tests, Pearson correlation, factorial analysis, structural equation modeling, and clusters. Since this investigation is subsidized by a contingent approach, it may be characterized as a systems\' research, relating the many variables under analysis like the noticed performance of contingent arrangements/fits. The presented results show an acceptability of the five hypotheses presented on the thesis, ratifying the environment influence on the intraorganizational contingent factors, the relations between the contingent factors and the company budgetary system, the contingent variables influence on the performance, the existent association between the budgetary attributes and the performance and, finally, fits/arrangements presence between the contingent variables and the budgetary system attributes, due to the better performance presentation. This empirical research reveals that unrelated to the company size, Paraná industries that show a higher degree of adherence to the budgetary system, industries that are conscious of the reasons for its uses and present routine tasks in its process showed an organizational and budgetary high performance. Thus, the conclusion is that under certain conditions the budgeting is a benefit to the company, revealing a positive cost-benefit relation.

A identificação de isolados dentro de algumas espécies de fungos filamentosos é de interesse médico, agrícola e industrial. Dentro do gênero Metarhizium, por exemplo, a identificação de espécies vêm sendo constantemente modificada devido a análise genética. Nesse contexto, o desenvolvimento de novas estratégias para a identificação rápida e confiável de microrganismos de maneira geral, e de fungos especificamente, é desejável. A espectroscopia no infravermelho por transformada de Fourier (FTIR) tornou-se amplamente difundida na medicina e biologia por fornecer rapidamente informações capazes de caracterizar bioquimicamente uma amostra. Assim, o objetivo desta pesquisa foi utilizar a espectroscopia de FTIR para a caracterização e classificação de linhagens e de espécies dos gêneros Aspergillus e Metarhizium, além da comparação entre diferentes metodologias para classificação. Foram selecionadas duas espécies do gênero Aspergillus: Aspergillus nidulans (ATCC 10074); Aspergillus flavus (ATCC 1410) com conídios verde e branco, obtidas da ATCC \"American Type Culture Collection\"(Manassas, VA), e seis espécies do gênero Metarhizium: Metarhizium acridum (ARSEF 324); Metarhizium acridum (ARSEF 3391); Metarhizium acridum (ARSEF 7486); Metarhizium anisopliae (ARSEF 5749); Metarhizium brunmeum (ARSEF 1095) e Metarhizium brunmeum (ARSEF 5626), obtidas da \"USDA-ARSEF collection of Entomopathogenic Fungal Cultures\"(U.S. Plant, Soil and Nutrition Laboratory, Ithaca, NY). O cultivo de todas as espécies foi feito em laboratório. As espécies A. nidulans ATCC 10074 e M. acridum ARSEF324 foram usadas para avaliar a influência da luz e do pH do meio de cultura sobre o crescimento dos fungos. Os valores de pH do meio estudados variaram ente 5 e 8. Para obter os espectros de absorção, os conídios foram delicadamente retirados e depositados sobre o cristal de ATR do espectrômetro do FTIR. Os espectros foram normalizados vetorialmente e subdivididos em quatro intervalos: 910-1178 cm-1 , 1178-1490 cm-1 , 1490-1790 cm-1 e 2810-2990 cm-1 . Parâmetros espectroscópicos como o deslocamento do pico de absorção e as razões entre áreas sob as bandas foram calculadas para todos os espectros coletados. O teste estatístico t-Student foi aplicado a estes parâmetros para diferenciar as amostras. Ambas as análises foram realizadas pelo programa Origin 8.5. A análise de componentes principais e agrupamento foram realizada usando o programa Minitab 16. As técnicas utilizadas para análise foram capazes de diferenciar as amostras de Aspergillus crescidas na presença e na ausência de luz, mas o mesmo não foi possível para o Metarhizium. Dentre os valores de pH do meio de cultura foi possível apenas a diferenciação do pH 8. O A. flavus com conídios brancos e verdes foram diferenciados entre si assim como entre eles e os A. nidulans. Entre as espécies foi possível obter a classificação de acordo com a taxonomia através da região espectral 1178-1490 cm-1. As quatro metodologias foram consideradas adequadas para análise dos espectros de absorção, entretanto o deslocamento de pico e a razão entre as áreas são viáveis apenas para um número pequeno de amostras enquanto o PCA e o cluster são viáveis para um número grande de amostras. / The identification of isolates within some species of filamentous fungi has medical, agricultural and industrial interest. Within the genus Metarhizium, for example, species identification is being constantly changed due to genetic analysis. In this context, the development of new strategies for the rapid and reliable identification of microorganisms in general, and specifically of fungi is desirable. Fourier Transform Infrared Spectroscopy (FTIR) has become widespread in medicine and biology by providing information quickly able to biochemically characterize a sample. The aim of this research was to use FTIR spectroscopy for characterization and classification of strains and species of the genera Aspergillus and Metarhizium and compare different methods for classification. We selected two strains of the genus Aspergillus: Aspergillus nidulans (ATCC 10074), Aspergillus flavus (ATCC 1410) with green and white conidia, obtained from ATCC \"American Type Culture Collection\"(Manassas, VA), and six strains of Metarhizium: Metarhizium acridum (ARSEF 324); Metarhizium acridum (ARSEF 3391); Metarhizium acridum (ARSEF 7486), Metarhizium anisopliae (ARSEF 5749); Metarhizium brunmeum (ARSEF 1095) and Metarhizium brunmeum (ARSEF 5626), obtained the \"USDA-ARSEF collection of Entomopathogenic Fungal Cultures \"(U.S. Plant, Soil and Nutrition Laboratory, Ithaca, NY). The cultivation of all species were made in the laboratory. The species A. nidulans (ATCC 10074) and M. acridum (ARSEF 324) were used to evaluate the influence of light and pH of the culture medium on the growth of fungi. The pH values of the medium studied ranged between 5 and 8. To obtain the absorption spectra, the conidia were gently removed and deposited on the ATR crystal of FTIR spectrometer. The absorption spectra were vector normalized and divided in four ranges: 910-1178 cm-1, 1178-1490cm-1, 1490-1790cm-1 and 2810-2990cm-1. Spectroscopic parameters as the displacement of the absorption peak and the ratios of the areas under the bands were calculated for all spectra collected. The statistical test t-Student was applied on these parameters to differentiate the samples. Both analyzes were performed by the program Origin 8.5. The principal component analysis and cluster analysis were performed using Minitab 16. The techniques used for analysis were able to differentiate samples of Aspergillus grown in the presence and absence of light, but it was not possible to Metarhizium. Among the pH of the culture medium was only possible to identify the pH 8. The A. flavus with white and green conidia were differentiated among themselves and between them and the A. nidulans. Among the strains of the same specie it was possible to obtain the correct classification according to the taxonomy through spectral region 1178-1490 cm-1. The four methods were considered appropriate for the analysis of absorption spectra, however displacement of the absorption peak and the ratios of the areas are feasible for a small number of samples and the multivariate analysis are feasible for a large number of samples.

Com a inclusão digital no Brasil, houve um grande crescimento não só do número de pessoas com acesso à Internet, como também do número de pessoas que realizam compras por esse meio. Esta investigação, cuja temática situa-se na área do Comportamento do Consumidor no ambiente online no momento da realização de uma compra, pretende construir um modelo de compreensão e verificação da propensão, ou predisposição, de um indivíduo para realizar compras online, a partir da análise das variáveis socioeconômicas e comportamentais. Por meio de análises estatísticas multivariadas - análise fatorial, análise de homogeneidade e análise de regressão logística -, concluiu-se que, entre os usuários da Internet pesquisados, há heterogeneidade na forma do comportamento de consumo e que existem diversas variáveis que influenciam o consumo por meio da Internet que dependem, em parte, de alguma característica, qualidade ou opinião singular. Este resultado aponta para a possibilidade de entendimento do processo de tomada de decisão pelos consumidores online, que facilitará as estratégias de marketing de forma a realizar uma captação de visitantes para os sites e, consequentemente, melhor conversão de visitantes em consumidores, quaisquer que sejam os produtos ou serviços. / The digital inclusion in Brazil has leveraged not only the number of individuals who access Internet, but also the amount of online shoppers. This essay, which delves into the area of Online Consumer Behavior at the moment of purchase, aims to design a model to verify and comprehend the tendency or inclination to shop online, regarding the analysis of social, economic and behavioral variables. Through multivariate analysis - factorial analysis, HOMALS and logistic regression -, it could be argued that, among the researched internet users, the web consumption behavior is influenced by an array of variables, resulting in a heterogenic pattern of e-shopping. The outcome points out a better understanding of online buyers\' decision making process, allowing marketing strategies to increase the number of online visitors and improve the conversion from visitors to consumers, whatever the product or service may be.

O presente trabalho propõe-se a analisar as afinidades biológicas através da comparação estatística multivariada dos dados craniométricos dos Paleoíndios de Lagoa Santa com outras séries esqueletais representativas de diversas regiões do mundo, testando, de forma complementar, as relações biológicas primeiramente observadas por Neves & Pucciarelli (1989). De acordo com a proposta inicial de Neves e Pucciarelli (1989), que sugere que a ocupação do Novo Mundo se deu a partir da chegada de dois componentes biológicos à América, os Paleoíndios, representantes da primeira leva de colonizadores, apresentam morfologia craniana generalizada, semelhante à observada entre as populações australo-melanésicas e africanas atuais e diametralmente oposta à morfologia especializada, característica dos nativos americanos recentes, todos descendentes da segunda leva de colonizadores. Para a execução do trabalho proposto foi utilizado, como base de dados comparativa à série Paleoíndia de Lagoa Santa, uma base de dados alternativa ao banco Howells, classicamente utilizado nas investigações bioantropologicas. A base de dados alternativa, numericamente expressiva e contando com populações originárias de 19 grandes regiões geográficas, foi organizada pelo Prof. Tsunehiko Hanihara, do Departamento de Anatomia da Saga Medical School (Saga, Japão), e cedida ao Laboratório de Estudos Evolutivos Humanos para a execução deste trabalho. Os resultados gerados através das técnicas estatísticas multivariadas empregadas indicaram que a população Paleoíndia de Lagoa Santa, em escala intracontinental, não se associa com as populações indígenas nativas americanas tardias e atuais e, em escala extra-continental, apresentam, de maneira geral, uma morfologia craniana mais semelhante à apresentada por populações de morfologia craniana generalizada / The present work proposes the analysis of the biological affinities, through a multivariate statistics comparison, of the Lagoa Santa Paleoindians´ craniometric data and other skeletal series, representatives of other regions in the world, testing, in a complementary way, the biological relationships first observed by Neves & Pucciarelli (1989). In accordance with Neves and Pucciarelli initial proposal, which suggests that the New World settlement happened with the arrive of two biological components to the America, the Paleoindians, representatives of the first migration wave, who show the generalized cranial morphology, resembling the one observed among the australomelanesian and african current populations, and completely opposite to the specialized morphology of the current native americans, all descendents from the second migration wave. For the accomplishment of this work it was used, as comparative data base to the Lagoa Santa\'s Paleoindians series, an alternative data set from that produced by Howells, commonly utilized in physical anthropological studies. The alternative data base, numerically expressive and counting with populations derived from 19 great geographic regions, was organized by the Professor Tsunehiko Hanihara, from the Anatomy Department of Saga Medical School (Saga, Japan), and granted to the Laboratory of Human Evolution Studies (Laboratório de Estudos Evolutivos Humanos - LEEH), for the execution of this project. The results generated through the employee multivariate statistics technics pointed out that the Paleoindian population from Lagoa Santa, in a intra-continental scale, do not associate with the american indigenous populations, neither from the past nor the present and, from the extra-continental scale, they present, in a general manner, a cranial morphology that resembles more that one presented by the populations with the generalized cranial morphology.

Com o objetivo de definir a composição química do rum a e encontrar descritores que permitam distinguir rums cubanos de rums não cubanos, foram analisadas 44 amostras de rum provenientes de 15 países diferentes. Essas análises foram realizadas por GC-FID, HPLC-UV-Vis, AAS e ICP OES. Os resultados analíticos foram analisados utilizando quimiometria (PLS, PCA, QDA, LDA, HCA e FA), que permitiu determinar quais analitos apresentam características significativas para separação dos rums em duas classes: Cubanos e Não Cubanos. / With the aim to define the chemical profile of rum and achieve descriptors that allows the distinction among Cubean rums and non- Cubean rums, the analysis of 44 samples of rum from 15 differents countries were reported. To provide the chemical descriptors, the analysis of mineral composition (cupper, iron, zinc, magnesium, manganese, cadmium, lead, nickel, cobalto, chromium, calcium, barium, sodium, lithium, strontium), phenolic profile (gallic acid, syringic acid, miricetin, ellagic acid, syringaldehyd, vanillin, vanillic acid, epicatechin, catechin, scopoletin, coniferaldehyde, sinapaldehyde, trans-resveratrol, quercetin, eugenol, coumarin), caramel content, alcohols composition (methanol, propanol, isobutanol, 2-butanol, butanol, isoamyl alcohol), acetic acid content, ethyl acetate content, aldehydes (formaldehyde, butyraldehyde, furfuraldehyde, 5-hydroxymethylfurfural, acetaldehyde, crotonaldehyde, propionaldehyde, isobutyraldehyde, benzaldehyde, valeraldehyde, isovaleraldehyde, acrolein), and ketones (2,3-butanodiona monoxime, ciclopentanone, acetilcetone, acetophenone, acetone, metil-isoamil-ketone) were carried out. The experimental results were theatead through exploratory and classification chemometric methods (principal component analysis, hierarchical cluster analysis, partial least squares, factor analysis, linear and quadratic discriminant analysis), that allowed determination of analytes which shown significant characteristics to separation rums into two distinct groups. The chemometric results highlighted that from all the classes of compounds studied were identified descriptors, with alternated participating loadings, that are capables to distinguish the spirit and classify it a acording to their origin. The PCA analysis present in this separation, an acumulative percentage in the first three principal components of 70.4%, having isoamyl alcohol and furfural as significant descriptors for PC1. In the classification of the Cuban rum and non –Cuban rum, the LDA model classified rums with 97.1% of accuracy using acetone, ethyl acetate, 2- butanol, 1-propanol, 1-butanol, isoamyl alcohol, caramel, barium, strontium, vanillic acid, vanillin, syringic acid, catechin, epicatechin, magnesium, acetaldehyde, ciclopentanonona, lead, isobutanol, zinc, calcium, sodium, and crotonaldehyde as descriptors, and with a inclusion of 10 blind samples to test the model, it was achived a percentage of correct assessment of 80% using those 23 chemical descriptors. These results suggested that the spirit, rum, show differences in its chemical composition related to mineral contents, aldehydes, ketones, alcohols, phenolic compounds, caramel, acetic acid and ethyl acetate, wich may be used to classify it according to the geographic origin.