Modelo de classificação para pequenas empresas supermercadistas quanto ao uso de técnicas de prevenção de perdas: um estudo de caso

The objective of this work is to draw attention to the importance of use of techniques of loss prevention in small retail organization, analyzing and creating a classification model related to the use of these in companies. This work identifies the fragilities and virtues of companies and classifies them relating the use of techniques of loss prevention. The used methodology is based in a revision of the available literature on measurements and techniques of loss prevention, analyzing the processes that techniques needed to be adopted to reduce losses, approaching the "pillars" of loss prevention, the cycle life of products in retail and cycles of continues improvement in business. Based on the objectives of this work and on the light of researched techniques, was defined the case study, developed from a questionnaire application and the researcher's observation on a net of 16 small supermarkets. From those studies a model of classification of companies was created. The practical implications of this work are useful to point mistakes in retail administration that can become losses, reducing the profitability of companies or even making them impracticable. The academic contribution of this study is a proposal of an unpublished model of classification for small supermarkets based on the use of techniques of loss prevention. As a result of the research, 14 companies were classified as Companies with Minimum Use of Loss Prevention Techniques - CMULPT, and 02 companies were classified as Companies with Deficient Use of Loss Prevention Techniques - CDULPT. The result of the research concludes that on average the group was classified as being Companies with Minimum Use of Techniques of Prevention of Losses EUMTPP, and that the companies should adopt a program of loss prevention focusing in the identification and quantification of losses and in a implantation of a culture of loss prevention

Diagnosis of inflammatory lesions by high-wavenumber FT-Raman spectroscopy

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Particle Competition and Cooperation in Networks for Semi-Supervised Learning

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

A wrapper approach for feature selection based on Bat Algorithm and Optimum-Path Forest

Besides optimizing classifier predictive performance and addressing the curse of the dimensionality problem, feature selection techniques support a classification model as simple as possible. In this paper, we present a wrapper feature selection approach based on Bat Algorithm (BA) and Optimum-Path Forest (OPF), in which we model the problem of feature selection as an binary-based optimization technique, guided by BA using the OPF accuracy over a validating set as the fitness function to be maximized. Moreover, we present a methodology to better estimate the quality of the reduced feature set. Experiments conducted over six public datasets demonstrated that the proposed approach provides statistically significant more compact sets and, in some cases, it can indeed improve the classification effectiveness. © 2013 Elsevier Ltd. All rights reserved.

Informação polarimétrica PALSAR/ALOS aplicada à discriminação de espécies e estimação de parâmetros morfológicos de macrófitas

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Avaliação de biomarcadores moleculares em mulheres com câncer de ovário

Pós-graduação em Ginecologia, Obstetrícia e Mastologia - FMB

Classificação de tábuas de madeira usando processamento de imagens digitais e aprendizado de máquina

Pós-graduação em Agronomia (Energia na Agricultura) - FCA

Classificação multicritério de peças de reposição

Pós-graduação em Engenharia Mecânica - FEG

Particle Competition and Cooperation in Networks for Semi-Supervised Learning

Semi-supervised learning is one of the important topics in machine learning, concerning with pattern classification where only a small subset of data is labeled. In this paper, a new network-based (or graph-based) semi-supervised classification model is proposed. It employs a combined random-greedy walk of particles, with competition and cooperation mechanisms, to propagate class labels to the whole network. Due to the competition mechanism, the proposed model has a local label spreading fashion, i.e., each particle only visits a portion of nodes potentially belonging to it, while it is not allowed to visit those nodes definitely occupied by particles of other classes. In this way, a "divide-and-conquer" effect is naturally embedded in the model. As a result, the proposed model can achieve a good classification rate while exhibiting low computational complexity order in comparison to other network-based semi-supervised algorithms. Computer simulations carried out for synthetic and real-world data sets provide a numeric quantification of the performance of the method.

Network-based stochastic semisupervised learning

Semisupervised learning is a machine learning approach that is able to employ both labeled and unlabeled samples in the training process. In this paper, we propose a semisupervised data classification model based on a combined random-preferential walk of particles in a network (graph) constructed from the input dataset. The particles of the same class cooperate among themselves, while the particles of different classes compete with each other to propagate class labels to the whole network. A rigorous model definition is provided via a nonlinear stochastic dynamical system and a mathematical analysis of its behavior is carried out. A numerical validation presented in this paper confirms the theoretical predictions. An interesting feature brought by the competitive-cooperative mechanism is that the proposed model can achieve good classification rates while exhibiting low computational complexity order in comparison to other network-based semisupervised algorithms. Computer simulations conducted on synthetic and real-world datasets reveal the effectiveness of the model.

On the impact of disproportional samples in credit scoring models: An application to a Brazilian bank data

Statistical methods have been widely employed to assess the capabilities of credit scoring classification models in order to reduce the risk of wrong decisions when granting credit facilities to clients. The predictive quality of a classification model can be evaluated based on measures such as sensitivity, specificity, predictive values, accuracy, correlation coefficients and information theoretical measures, such as relative entropy and mutual information. In this paper we analyze the performance of a naive logistic regression model (Hosmer & Lemeshow, 1989) and a logistic regression with state-dependent sample selection model (Cramer, 2004) applied to simulated data. Also, as a case study, the methodology is illustrated on a data set extracted from a Brazilian bank portfolio. Our simulation results so far revealed that there is no statistically significant difference in terms of predictive capacity between the naive logistic regression models and the logistic regression with state-dependent sample selection models. However, there is strong difference between the distributions of the estimated default probabilities from these two statistical modeling techniques, with the naive logistic regression models always underestimating such probabilities, particularly in the presence of balanced samples. (C) 2012 Elsevier Ltd. All rights reserved.

Sexuality and the ‘Relations of Ruling’: Using Institutional Ethnography to Research Lesbian and Gay Foster Care and Adoption.

This article makes use of institutional ethnography to research foster care and adoption by lesbians and gay men, drawing on the work of the feminist sociologist Dorothy E. Smith in order to demonstrate the investigation of social work institutional categories and the ‘relations of ruling’. Through an analysis of the ways in which ‘gender’ and the idea of the ‘gender role model’ is used within the assessment of gay and lesbian foster carers and adopters, the author shows how these categories are produced and used to police relationship forms and to identify ‘deviant instances’.

Metodología de evaluación de la seguridad vial en intersecciones basada en el análisis cuantitativo de conflictos entre vehículos

El modo tradicional de estimar el nivel de seguridad vial es el registro de accidentes de tráfico, sin embargo son altamente variables, aleatorios y necesitan un periodo de registro de al menos 3 años. Existen metodologías preventivas en las cuales no es necesario que ocurra un accidente para determinar el nivel de seguridad de una intersección, como lo es la técnica de los conflictos de tráfico, que introduce las mediciones alternativas de seguridad como cuantificadoras del riesgo de accidente. El objetivo general de la tesis es establecer una metodología que permita clasificar el riesgo en intersecciones interurbanas, en función del análisis de conflictos entre vehículos, realizado mediante las variables alternativas o indirectas de seguridad vial. La metodología para el análisis y evaluación temprana de la seguridad en una intersección, estará basada en dos medidas alternativas de seguridad: el tiempo hasta la colisión y el tiempo posterior a la invasión de la trayectoria. El desarrollo experimental se realizó mediante estudios de campo, para la parte exploratoria de la investigación, se seleccionaron 3 intersecciones interurbanas en forma de T donde se obtuvieron las variables que caracterizan los conflictos entre vehículos; luego mediante técnicas de análisis multivariante, se obtuvo los modelos de clasificación del riesgo cualitativo y cuantitativo. Para la homologación y el estudio final de concordancia entre el índice propuesto y el modelo de clasificación, se desarrollaron nuevos estudios de campo en 6 intersecciones interurbanas en forma de T. El índice de riesgo obtenido resulta una herramienta muy útil para realizar evaluaciones rápidas conducentes a estimar la peligrosidad de una intersección en T, debido a lo simple y económico que resulta obtener los registros de datos en campo, por medio de una rápida capacitación a operarios; la elaboración del informe de resultados debe ser por un especialista. Los índices de riesgo obtenidos muestran que las variables originales más influyentes son las mediciones de tiempo. Se pudo determinar que los valores más altos del índice de riesgo están relacionados a un mayor riesgo de que un conflicto termine en accidente. Dentro de este índice, la única variable cuyo aporte es proporcionalmente directo es la velocidad de aproximación, lo que concuerda con lo que sucede en un conflicto, pues una velocidad excesiva se manifiesta como un claro factor de riesgo ya que potencia todos los fallos humanos en la conducción. Una de las principales aportaciones de esta tesis doctoral a la ingeniería de carreteras, es la posibilidad de aplicación de la metodología por parte de administraciones de carreteras locales, las cuales muchas veces cuentan con recursos de inversión limitados para efectuar estudios preventivos, sobretodo en países en vías de desarrollo. La evaluación del riesgo de una intersección luego de una mejora en cuanto a infraestructura y/o dispositivos de control de tráfico, al igual que un análisis antes – después, pero sin realizar una comparación mediante la ocurrencia de accidentes, sino que por medio de la técnica de conflictos de tráfico, se puede convertir en una aplicación directa y económica. Además, se pudo comprobar que el análisis de componentes principales utilizado en la creación del índice de riesgo de la intersección, es una herramienta útil para resumir todo el conjunto de mediciones que son posibles de obtener con la técnica de conflictos de tráfico y que permiten el diagnóstico del riesgo de accidentalidad en una intersección. En cuanto a la metodología para la homologación de los modelos, se pudo establecer la validez y confiabilidad al conjunto de respuestas entregadas por los observadores en el registro de datos en campo, ya que los resultados de la validación establecen que la medición de concordancia de las respuestas entregadas por los modelos y lo observado, son significativas y sugieren una alta coincidencia entre ellos. ABSTRACT The traditional way of estimating road safety level is the record of occurrence of traffic accidents; however, they are highly variable, random, and require a recording period of at least three years. There are preventive methods which do not need an accident to determine the road safety level of an intersection, such as traffic conflict technique, which introduces surrogate safety measures as parameters for the evaluation of accident risks. The general objective of the thesis is to establish a methodology that will allow the classification of risk at interurban intersections as a function of the analysis of conflicts between vehicles performed by means of surrogate road safety variables. The proposal of a methodology for the analysis and early evaluation of safety at an intersection will be based on two surrogate safety measures: the time to collision and the post encroachment time. On the other hand, the experimental development has taken place by means of field studies in which the exploratory part of the investigation selected three interurban T-intersections where the application of the traffic conflict technique gave variables that characterize the conflicts between vehicles; then, using multivariate analysis techniques, the models for the classification of qualitative and quantitative risk were obtained. With the models new field studies were carried out at six interurban Tintersections with the purpose of developing the homologation and the final study of the agreement between the proposed index and the classification model. The risk index obtained is a very useful tool for making rapid evaluations to estimate the hazard of a T-intersection, as well as for getting simply and economically the field data records after a fast training of the workers and then preparing the report of results by a specialist. The risk indices obtained show that the most influential original variables are the measurements of time. It was determined that the highest risk index values are related with greater risk of a conflict resulting in an accident. Within this index, the only variable whose contribution is proportionally direct is the approach speed, in agreement with what happens in a conflict, because excessive speed appears as a clear risk factor at an intersection because it intensifies all the human driving faults. One of the main contributions of this doctoral thesis to road engineering is the possibility of applying the methodology by local road administrations, which very often have limited investment resources to carry out these kinds of preventive studies, particularly in developing countries. The evaluation of the risk at an intersection after an improvement in terms of infrastructure and/or traffic control devices, the same as a before/after analysis, without comparison of accident occurrence but by means of the traffic conflict technique, can become a direct and economical application. It is also shown that main components analysis used for producing the risk index of the intersection is a useful tool for summarizing the whole set of measurements that can be obtained with the traffic conflict technique and allow diagnosing accident risk at an intersection. As to the methodology for the homologation of the models, the validity and reliability of the set of responses delivered by the observers recording the field data could be established, because the results of the validation show that agreement between the observations and the responses delivered by the models is significant and highly coincident.

Self-configuring data mining for ubiquitous computing

Ubiquitous computing software needs to be autonomous so that essential decisions such as how to configure its particular execution are self-determined. Moreover, data mining serves an important role for ubiquitous computing by providing intelligence to several types of ubiquitous computing applications. Thus, automating ubiquitous data mining is also crucial. We focus on the problem of automatically configuring the execution of a ubiquitous data mining algorithm. In our solution, we generate configuration decisions in a resource aware and context aware manner since the algorithm executes in an environment in which the context often changes and computing resources are often severely limited. We propose to analyze the execution behavior of the data mining algorithm by mining its past executions. By doing so, we discover the effects of resource and context states as well as parameter settings on the data mining quality. We argue that a classification model is appropriate for predicting the behavior of an algorithm?s execution and we concentrate on decision tree classifier. We also define taxonomy on data mining quality so that tradeoff between prediction accuracy and classification specificity of each behavior model that classifies by a different abstraction of quality, is scored for model selection. Behavior model constituents and class label transformations are formally defined and experimental validation of the proposed approach is also performed.

Mars: a personalised mobile activity recognition system

Mobile activity recognition focuses on inferring the current activities of a mobile user by leveraging the sensory data that is available on today’s smart phones. The state of the art in mobile activity recognition uses traditional classification learning techniques. Thus, the learning process typically involves: i) collection of labelled sensory data that is transferred and collated in a centralised repository; ii) model building where the classification model is trained and tested using the collected data; iii) a model deployment stage where the learnt model is deployed on-board a mobile device for identifying activities based on new sensory data. In this paper, we demonstrate the Mobile Activity Recognition System (MARS) where for the first time the model is built and continuously updated on-board the mobile device itself using data stream mining. The advantages of the on-board approach are that it allows model personalisation and increased privacy as the data is not sent to any external site. Furthermore, when the user or its activity profile changes MARS enables promptly adaptation. MARS has been implemented on the Android platform to demonstrate that it can achieve accurate mobile activity recognition. Moreover, we can show in practise that MARS quickly adapts to user profile changes while at the same time being scalable and efficient in terms of consumption of the device resources.