Using a Bayesian model for bankruptcy prediction : a comparative approach

The purpose of this study is to examine the impact of the choice of cut-off points, sampling procedures, and the business cycle on the accuracy of bankruptcy prediction models. Misclassification can result in erroneous predictions leading to prohibitive costs to firms, investors and the economy. To test the impact of the choice of cut-off points and sampling procedures, three bankruptcy prediction models are assessed- Bayesian, Hazard and Mixed Logit. A salient feature of the study is that the analysis includes both parametric and nonparametric bankruptcy prediction models. A sample of firms from Lynn M. LoPucki Bankruptcy Research Database in the U. S. was used to evaluate the relative performance of the three models. The choice of a cut-off point and sampling procedures were found to affect the rankings of the various models. In general, the results indicate that the empirical cut-off point estimated from the training sample resulted in the lowest misclassification costs for all three models. Although the Hazard and Mixed Logit models resulted in lower costs of misclassification in the randomly selected samples, the Mixed Logit model did not perform as well across varying business-cycles. In general, the Hazard model has the highest predictive power. However, the higher predictive power of the Bayesian model, when the ratio of the cost of Type I errors to the cost of Type II errors is high, is relatively consistent across all sampling methods. Such an advantage of the Bayesian model may make it more attractive in the current economic environment. This study extends recent research comparing the performance of bankruptcy prediction models by identifying under what conditions a model performs better. It also allays a range of user groups, including auditors, shareholders, employees, suppliers, rating agencies, and creditors' concerns with respect to assessing failure risk.

Échantillonnage et modélisation de l’habitat des communautés de poissons de rivière des basses Laurentides

Plusieurs études à grande échelle ont identifié la modification ou la perte d’habitats comme menace principale à la conservation des communautés de poissons d’eau douce. Au Canada, « aucune perte nette dans la capacité productive des habitats » (NNL) est le principe directeur de la politique de gestion des habitats du ministère des Pêches et Océans. Le respect du NNL implique l’avancement des connaissances au niveau des relations entre les poissons et leurs habitats, de même que des outils pour quantifier l’impact de la modification des habitats sur les poissons. Les modèles d’utilisation de l’habitat des poissons (FHUM) sont des outils qui permettent d’améliorer nos connaissances des relations poissons – habitat, de prédire la distribution des espèces, mais aussi leurs densités, biomasses ou abondances, sur la base des caractéristiques de l’environnement. L’objectif général de mon mémoire est d’améliorer la performance des FHUM pour les rivières des basses Laurentides, en suggérant des perfectionnements au niveau de 2 aspects cruciaux de l’élaboration de tels modèles : la description précise de la communauté de poissons et l’utilisation de modèles statistiques efficaces. Dans un premier chapitre, j’évalue la performance relative de la pêcheuse électrique et de l’échantillonnage en visuel (plongée de surface) pour estimer les abondances des combinaisons d’espèces et de classes de taille des poissons en rivière. J’évalue aussi l’effet des conditions environnementales sur les différences potentielles entre les communautés observées par ces 2 méthodes d’échantillonnage. Pour ce faire, 10 sections de rivière de 20 m de longueur ont été échantillonnées à l’aide de ces 2 méthodes alors qu’elles étaient fermées par des filets de blocage. 3 plongeurs performèrent l’échantillonnage en visuel en se déplaçant de l’aval vers l’amont des sections, tout en dénombrant les espèces et classes de taille. Par la suite, nous avons fait 3 passages de pêcheuse électrique et les abondances furent estimées grâce à un modèle restreint de maximum de vraisemblance, basé sur la diminution des abondances observées. De plus grandes abondances de poissons furent observées en visuel qu’avec la pêcheuse électrique à tous les sites. La richesse spécifique observée en visuel était plus élevée (6/10) ou égale (4/10) à celle observée avec la pêcheuse électrique. Les différences entre les communautés de poissons observées à l’aide de ces 2 méthodes ne purent être reliées aux conditions environnementales. Les résultats de cette expérience sont contraires à ceux de toutes les études comparant ces 2 méthodes d’échantillonnage, lesquels suggèrent une supériorité de la pêcheuse électrique. Les conditions environnementales de notre expérience étaient distinctes de celles observées dans les autres études (absence d’arbres tombés dans l’eau, très peu de substrats grossiers), mais la différence la plus marquante était en terme de communauté de poissons observée (dominance des cyprinidés et des centrarchidés plutôt que des salmonidés). Je termine ce chapitre en suggérant que les caractéristiques comportementales favorisant l’évitement de la capture (formation de bancs) et facilitant l’observation en visuel (curiosité) sont responsables de la supériorité de la plongée de surface pour échantillonner les communautés dans les rivières des basses Laurentides. Dans un deuxième chapitre, je développe des FHUM pour des communautés de poissons de rivière ayant plusieurs espèces. Dans le but de simplifier la modélisation de telles communautés et d’améliorer notre compréhension des relations poissons – habitat, j’utilise les concepts de guilde écologique et de filtre environnemental pour explorer les relations entre les guildes formées sur la bases de différents types de traits (reproducteurs, taxonomiques, éco-morphologiques et alimentaires) et les conditions environnementales locales à l’échelle du méso-habitat. Les modèles d’habitats basés sur les guildes reproductrices ont clairement surpassé les autres modèles, parce que l’habitat de fraie reflète l’habitat de préférence en dehors de la période de reproduction. J’ai également utilisé l’approche inverse, c’est à dire définir des guildes d’utilisation de l’habitat et les mettre en relation avec les traits des espèces. Les traits reliés à l’alimentation des poissons ont semblés être les meilleurs pour expliquer l’appartenance aux groupes d’utilisation de l’habitat, mais le modèle utilisé ne représentait pas bien la relation entre les groupes. La validation de notre modèle basé sur les guildes reproductrices avec un jeu de données indépendant pourrait confirmer notre découverte, laquelle représente une manière prometteuse de modéliser les relations poissons – environnement dans des communautés de poissons complexes. En conclusion, mon mémoire suggère d’importantes améliorations aux FHUM pour les communautés de poissons des basses Laurentides, en suggérant de prendre en compte les caractéristiques biologiques des cours d’eau dans le choix d’une méthode d’échantillonnage, et également en utilisant une méthode prometteuse pour simplifier les FHUM de communautés de poissons complexes : les guildes reproductrices.

Développement d’une approche quantitative pour l’étude du poumon équin : fixation et échantillonnage pour l’application des principes de la stéréologie

La présente étude visait à développer un protocole de fixation et d'échantillonnage pour le poumon équin suivant les directives publiées sur l’utilisation d’une approche stéréologique de type « design-based ». Les poumons gauches de chevaux contrôles et atteints du souffle ont été fixés avec du formaldéhyde 10% pendant 48h à une pression constante de 25-30 cm d’H2O. Les poumons ont été sectionnés en 20-21 tranches d’une épaisseur d'environ 2,5 cm chacune; de 10-11 tranches ont été sélectionnées de façon aléatoire et systématique pour la mesure du volume de référence avec la méthode de Cavalieri. Un protocole d’échantillonnage systématique, aléatoire et uniforme utilisant le principe du « smooth fractionator » et un poinçon à biopsie de 17 mm ont été utilisés pour échantillonner une fraction représentative de chaque poumon. Les méthodes d’échantillonnage de sections verticales, uniformes et aléatoires (VUR) et d’échantillonnage isotropique, uniforme et aléatoire (IUR) ont toutes deux été effectuées pour comparer le nombre de voies respiratoires en coupe perpendiculaire obtenues à partir de chaque méthode. L'architecture globale et la qualité des tissus fixés ont également été évaluées. Des spécimens pulmonaires équins ont été échantillonnés avec succès selon un protocole visant à produire des données morphométriques valides. Les tissus ont été fixés avec un minimum d'artéfacts et contenaient une quantité suffisante de voies respiratoires en coupe perpendiculaire dans les deux types d’échantillons. En conclusion, un protocole de fixation et d'échantillonnage adapté au poumon équin permettant l'utilisation d'une approche stéréologique de type « design-based » a été élaboré pour l’étude du remodelage des voies respiratoires.

Contributions of local, lateral and contextual habitat variables to explaining variation in fisheries productivity metrics in the littoral zone of a reservoir

Puisque l’altération des habitats d’eau douce augmente, il devient critique d’identifier les composantes de l’habitat qui influencent les métriques de la productivité des pêcheries. Nous avons comparé la contribution relative de trois types de variables d’habitat à l’explication de la variance de métriques d’abondance, de biomasse et de richesse à l’aide de modèles d’habitat de poissons, et avons identifié les variables d’habitat les plus efficaces à expliquer ces variations. Au cours des étés 2012 et 2013, les communautés de poissons de 43 sites littoraux ont été échantillonnées dans le Lac du Bonnet, un réservoir dans le Sud-est du Manitoba (Canada). Sept scénarios d’échantillonnage, différant par l’engin de pêche, l’année et le moment de la journée, ont été utilisés pour estimer l’abondance, la biomasse et la richesse à chaque site, toutes espèces confondues. Trois types de variables d’habitat ont été évalués: des variables locales (à l’intérieur du site), des variables latérales (caractérisation de la berge) et des variables contextuelles (position relative à des attributs du paysage). Les variables d’habitat locales et contextuelles expliquaient en moyenne un total de 44 % (R2 ajusté) de la variation des métriques de la productivité des pêcheries, alors que les variables d’habitat latérales expliquaient seulement 2 % de la variation. Les variables les plus souvent significatives sont la couverture de macrophytes, la distance aux tributaires d’une largeur ≥ 50 m et la distance aux marais d’une superficie ≥ 100 000 m2, ce qui suggère que ces variables sont les plus efficaces à expliquer la variation des métriques de la productivité des pêcheries dans la zone littorale des réservoirs.

Evaluation of major feed resources in crop-livestock mixed farming systems, southern Ethiopia: Indigenous knowledge versus laboratory analysis results

Evaluation of major feed resources was conducted in four crop-livestock mixed farming systems of central southern Ethiopia, with 90 farmers, selected using multi-stage purposive and random sampling methods. Discussions were held with focused groups and key informants for vernacular name identification of feed, followed by feed sampling to analyse chemical composition (CP, ADF and NDF), in-vitro dry matter digestibility (IVDMD), and correlate with indigenous technical knowledge (ITK). Native pastures, crop residues (CR) and multi-purpose trees (MPT) are the major feed resources, demonstrated great variations in seasonality, chemical composition and IVDMD. The average CP, NDF and IVDMD values for grasses were 83.8 (ranged: 62.9–190), 619 (ranged: 357–877) and 572 (ranged: 317–743) g kg^(−1) DM, respectively. Likewise, the average CP, NDF and IVDMD for CR were 58 (ranged: 20–90), 760 (ranged: 340–931) and 461 (ranged: 285–637)g kg^(−1) DM, respectively. Generally, the MPT and non-conventional feeds (NCF, Ensete ventricosum and Ipomoea batatas) possessed higher CP (ranged: 155–164 g kg^(−1) DM) and IVDMD values (611–657 g kg^(−1) DM) while lower NDF (331–387 g kg^(−1) DM) and ADF (321–344 g kg^(−1) DM) values. The MPT and NCF were ranked as the best nutritious feeds by ITK while crop residues were the least. This study indicates that there are remarkable variations within and among forage resources in terms of chemical composition. There were also complementarities between ITK and feed laboratory results, and thus the ITK need to be taken into consideration in evaluation of local feed resources.

Experimental verification of suction sampler capture efficiency in grasslands of differing vegetation height and structure

1. Suction sampling is a popular method for the collection of quantitative data on grassland invertebrate populations, although there have been no detailed studies into the effectiveness of the method. 2. We investigate the effect of effort (duration and number of suction samples) and sward height on the efficiency of suction sampling of grassland beetle, true bug, planthopper and spider Populations. We also compare Suction sampling with an absolute sampling method based on the destructive removal of turfs. 3. Sampling for durations of 16 seconds was sufficient to collect 90% of all individuals and species of grassland beetles, with less time required for the true bugs, spiders and planthoppers. The number of samples required to collect 90% of the species was more variable, although in general 55 sub-samples was sufficient for all groups, except the true bugs. Increasing sward height had a negative effect on the capture efficiency of suction sampling. 4. The assemblage structure of beetles, planthoppers and spiders was independent of the sampling method (suction or absolute) used. 5. Synthesis and applications. In contrast to other sampling methods used in grassland habitats (e.g. sweep netting or pitfall trapping), suction sampling is an effective quantitative tool for the measurement of invertebrate diversity and assemblage structure providing sward height is included as a covariate. The effective sampling of beetles, true bugs, planthoppers and spiders altogether requires a minimum sampling effort of 110 sub-samples of duration of 16 seconds. Such sampling intensities can be adjusted depending on the taxa sampled, and we provide information to minimize sampling problems associated with this versatile technique. Suction sampling should remain an important component in the toolbox of experimental techniques used during both experimental and management sampling regimes within agroecosystems, grasslands or other low-lying vegetation types.

Measuring bee diversity in different European habitats and biogeographical regions

Bee pollinators are currently recorded with many different sampling methods. However, the relative performances of these methods have not been systematically evaluated and compared. In response to the strong need to record ongoing shifts in pollinator diversity and abundance, global and regional pollinator initiatives must adopt standardized sampling protocols when developing large-scale and long-term monitoring schemes. We systematically evaluated the performance of six sampling methods (observation plots, pan traps, standardized and variable transect walks, trap nests with reed internodes or paper tubes) that are commonly used across a wide range of geographical regions in Europe and in two habitat types (agricultural and seminatural). We focused on bees since they represent the most important pollinator group worldwide. Several characteristics of the methods were considered in order to evaluate their performance in assessing bee diversity: sample coverage, observed species richness, species richness estimators, collector biases (identified by subunit-based rarefaction curves), species composition of the samples, and the indication of overall bee species richness (estimated from combined total samples). The most efficient method in all geographical regions, in both the agricultural and seminatural habitats, was the pan trap method. It had the highest sample coverage, collected the highest number of species, showed negligible collector bias, detected similar species as the transect methods, and was the best indicator of overall bee species richness. The transect methods were also relatively efficient, but they had a significant collector bias. The observation plots showed poor performance. As trap nests are restricted to cavity-nesting bee species, they had a naturally low sample coverage. However, both trap nest types detected additional species that were not recorded by any of the other methods. For large-scale and long-term monitoring schemes with surveyors with different experience levels, we recommend pan traps as the most efficient, unbiased, and cost-effective method for sampling bee diversity. Trap nests with reed internodes could be used as a complementary sampling method to maximize the numbers of collected species. Transect walks are the principal method for detailed studies focusing on plant-pollinator associations. Moreover, they can be used in monitoring schemes after training the surveyors to standardize their collection skills.

Epidemic K-Means clustering

The K-Means algorithm for cluster analysis is one of the most influential and popular data mining methods. Its straightforward parallel formulation is well suited for distributed memory systems with reliable interconnection networks. However, in large-scale geographically distributed systems the straightforward parallel algorithm can be rendered useless by a single communication failure or high latency in communication paths. This work proposes a fully decentralised algorithm (Epidemic K-Means) which does not require global communication and is intrinsically fault tolerant. The proposed distributed K-Means algorithm provides a clustering solution which can approximate the solution of an ideal centralised algorithm over the aggregated data as closely as desired. A comparative performance analysis is carried out against the state of the art distributed K-Means algorithms based on sampling methods. The experimental analysis confirms that the proposed algorithm is a practical and accurate distributed K-Means implementation for networked systems of very large and extreme scale.

Fault tolerant decentralised K-Means clustering for asynchronous large-scale networks

The K-Means algorithm for cluster analysis is one of the most influential and popular data mining methods. Its straightforward parallel formulation is well suited for distributed memory systems with reliable interconnection networks, such as massively parallel processors and clusters of workstations. However, in large-scale geographically distributed systems the straightforward parallel algorithm can be rendered useless by a single communication failure or high latency in communication paths. The lack of scalable and fault tolerant global communication and synchronisation methods in large-scale systems has hindered the adoption of the K-Means algorithm for applications in large networked systems such as wireless sensor networks, peer-to-peer systems and mobile ad hoc networks. This work proposes a fully distributed K-Means algorithm (EpidemicK-Means) which does not require global communication and is intrinsically fault tolerant. The proposed distributed K-Means algorithm provides a clustering solution which can approximate the solution of an ideal centralised algorithm over the aggregated data as closely as desired. A comparative performance analysis is carried out against the state of the art sampling methods and shows that the proposed method overcomes the limitations of the sampling-based approaches for skewed clusters distributions. The experimental analysis confirms that the proposed algorithm is very accurate and fault tolerant under unreliable network conditions (message loss and node failures) and is suitable for asynchronous networks of very large and extreme scale.

The utility of aerial pan-trapping for assessing insect pollinators across vertical strata

Insect pollinators provide a critical ecosystem service by pollinating many wild flowers and crops. It is therefore essential to be able to effectively survey and monitor pollinator communities across a range of habitats, and in particular, sample the often stratified parts of the habitats where insects are found. To date, a wide array of sampling methods have been used to collect insect pollinators, but no single method has been used effectively to sample across habitat types and throughout the spatial structure of habitats. Here we present a method of ‘aerial pan-trapping’ that allows insect pollinators to be sampled across the vertical strata from the canopy of forests to agro-ecosystems. We surveyed and compared the species richness and abundance of a wide range of insect pollinators in agricultural, secondary regenerating forest and primary forest habitats in Ghana to evaluate the usefulness of this approach. In addition to confirming the efficacy of the method at heights of up to 30 metres and the effects of trap color on catch, we found greatest insect abundance in agricultural land and higher bee abundance and species richness in undisturbed forest compared to secondary forest.

Estimating abundance and spatial distribution patterns of the bubble crab Dotilla fenestrata (Crustacea : Brachyura)

The bubble crab Dotilla fenestrata forms very dense populations on the sand flats of the eastern coast of Inhaca Island, Mozambique, making it an interesting biological model to examine spatial distribution patterns and test the relative efficiency of common sampling methods. Due to its apparent ecological importance within the sandy intertidal community, understanding the factors ruling the dynamics of Dotilla populations is also a key issue. In this study, different techniques of estimating crab density are described, and the trends of spatial distribution of the different population categories are shown. The studied populations are arranged in discrete patches located at the well-drained crests of nearly parallel mega sand ripples. For a given sample size, there was an obvious gain in precision by using a stratified random sampling technique, considering discrete patches as strata, compared to the simple random design. Density average and variance differed considerably among patches since juveniles and ovigerous females were found clumped, with higher densities at the lower and upper shore levels, respectively. Burrow counting was found to be an adequate method for large-scale sampling, although consistently underestimating actual crab density by nearly half. Regression analyses suggested that crabs smaller than 2.9 mm carapace width tend to be undetected in visual burrow counts. A visual survey of sampling plots over several patches of a large Dotilla population showed that crab density varied in an interesting oscillating pattern, apparently following the topography of the sand flat. Patches extending to the lower shore contained higher densities than those mostly covering the higher shore. Within-patch density variability also pointed to the same trend, but the density increment towards the lowest shore level varied greatly among the patches compared.

Spatial analysis of the distribution of Spodoptera frugiperda (J.E. Smith) (Lepidoptera: Noctuidae) and losses in maize crop productivity using geostatistics

A lagarta-do-cartucho, Spodoptera frugiperda (J.E. Smith), é uma das principais pragas do milho nas Américas. O estudo de sua distribuição espacial é fundamental para a utilização de estratégias de controle, otimização de técnicas de amostragens, determinação de danos econômicos e incorporação de um programa de agricultura de precisão. em uma área cultivada com milho foram realizadas amostragens com intervalo semanal, correspondendo ao estádio vegetativo que compreende desde a germinação até o pendoamento. Foram amostradas 10 plantas ao acaso por parcela, no total de 2000 plantas em cada amostragem. A produtividade foi obtida através da colheita de todas as parcelas que eram pesadas separadamente no campo e em cada parcela foram coletadas 15 espigas aleatoriamente para estimar o comprimento e o diâmetro médio. As análises espaciais, utilizando geoestatística, mostraram que o modelo esférico apresentou o melhor ajuste às lagartas pequenas. À medida que as lagartas foram se desenvolvendo sua distribuição foi tornando aleatória, representada por um modelo ajustado por uma reta, não tendo sido detectado nenhum tipo de dependência espacial nos pontos de amostragem. A produtividade e o diâmetro e comprimento da espiga foram descritos por modelos esféricos, indicando uma variabilidade espacial nos parâmetros de produtividade na área cultivada. A geoestatística mostrou-se promissora para a aplicação de métodos precisos no controle integrado de pragas.

FLUTUAÇÃO POPULACIONAL DE Dione juno juno (Cramer, 1779) (Lepidoptera: Nymphalidae) em MARACUJAZEIROS (Passiflora spp.), MÉTODOS DE AMOSTRAGEM E RESISTÊNCIA DE GENÓTIPOS

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Avaliação da ocorrência sazonal de percevejos em cinco espécies de maracujazeiro, utilizando dois métodos de amostragem

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Amostragem, diversidade e sazonalidade de Hemerobiidae (Neuroptera) em Coffea arabica L. cv. Obatã (Rubiaceae)

Este estudo teve por objetivo avaliar métodos de amostragem, abundância sazonal e diversidade da população de Hemerobiidae associada a cultivo de café Coffea arabica L. cv. Obatã em Cravinhos, São Paulo, Brasil. Para tanto foram realizadas amostragens semanais no período de maio de 2005 a abril de 2006. Os métodos de amostragem utilizados foram: rede de varredura e armadilhas de Möericke e luminosa. Foram coletados 491 exemplares de Hemerobiidae pertencentes a quatro gêneros: Nusalala (231 espécimes / 47,2% do total de hemerobiídeos coletados), Megalomus (110 / 22,5%), Hemerobius (104 / 21,3%) e Sympherobius (44 / 9%). A rede de varredura foi a mais eficiente para a captura de Hemerobiidae e a armadilha de Möericke foi o método de amostragem que apresentou os maiores valores de diversidade (H'= 0,56) e de equitabilidade (J= 0,93). Os hemerobiídeos estiveram presentes na área estudada durante o ano todo; as maiores freqüências foram registradas entre agosto e março (final do inverno, primavera e verão) e o maior pico populacional ocorreu em janeiro (na metade do verão). Megalomus apresentou correlação positiva e significativa (p< 0,05) com a precipitação pluviométrica e as temperaturas máxima e mínima; Nusalala com as temperaturas máxima e mínima e, Sympherobius apenas com a temperatura máxima.