Your browser doesn't support javascript.
loading
Mostrar: 20 | 50 | 100
Resultados 1 - 6 de 6
Filtrar
Mais filtros










Intervalo de ano de publicação
1.
An. psicol ; 30(2): 633-641, mayo 2014. ilus, tab
Artigo em Inglês | IBECS | ID: ibc-121802

RESUMO

El presente trabajo tiene el propósito de analizar el poder predictivo de diversas variables psicosociales y de personalidad sobre el con-sumo o no consumo de nicotina en la población adolescente mediante el uso de diversas técnicas de clasificación procedentes de la metodología Data Mining. Más concretamente, se analizan las RNA -Perceptrón Multicapa (MLP), Funciones de Base Radial (RBF) y Redes Probabilísticas (PNN)--, los árboles de decisión, el modelo de regresión logística y el análisis discriminante. Para ello, se ha trabajado con una muestra de 2666 adolescentes, de los cuales 1378 no consumen nicotina mientras que 1288 son consumidores de nicotina. Los modelos analizados han sido capaces de discriminar correctamente entre ambos tipos de sujeto en un rango comprendido entre el 77.39% y el 78.20%, alcanzando una sensibilidad del 91.29% y una especificidad del 74.32%. Con este estudio, se pone a disposición del especialista en conductas adictivas, un conjunto de técnicas estadísticas avanzadas capaces de manejar simultáneamente una gran cantidad de variables y sujetos, así como aprender de forma automática patrones y relaciones complejas, siendo muy adecuadas para la predicción y prevención del comporta-miento adictivo


This study is aimed at analysing the predictive power of different psychosocial and personality variables on the consumption or non-consumption of nicotine in a teenage population using different classification techniques from the field of Data Mining. More specifically, we analyse ANNs - Multilayer Perceptron (MLP), Radial Basis Functions (RBF) and Probabilistic Neural Networks (PNNs) -decision trees, the logistic regression model and discriminant analysis. To this end, we worked with a sample of 2666 teenagers, 1378 of whom do not consume nicotine while 1288 are nicotine consumers. The models analysed were able to discriminate correctly between both types of subjects within a range of 77.39% to 78.20%, achieving 91.29% sensitivity and 74.32% specificity. With this study, we place at the disposal of specialists in addictive behaviours a set of advanced statistical techniques that are capable of simultaneously processing a large quantity of variables and subjects, as well as learning complex patterns and relationships automatically, in such a way that they are very appropriate for predicting and preventing addictive behaviour


Assuntos
Humanos , Masculino , Feminino , Adolescente , Fumar/psicologia , Psicometria/instrumentação , Comportamento Aditivo/psicologia , Comportamento do Adolescente/psicologia , Fatores de Risco , Risco Ajustado/métodos , Modelos Logísticos
2.
Psicothema (Oviedo) ; 25(4): 500-506, oct.-dic. 2013. ilus, tab
Artigo em Inglês | IBECS | ID: ibc-115898

RESUMO

Background: The mean absolute percentage error (MAPE) is probably the most widely used goodness-of-fit measure. However, it does not meet the validity criterion due to the fact that the distribution of the absolute percentage errors is usually skewed to the right, with the presence of outlier values. In these cases, MAPE overstates the corresponding population parameter. In this study, we propose an alternative index, called Resistant MAPE or R-MAPE based on the calculation of the Huber M-estimator, which allows overcoming the aforementioned limitation. Method: The results derived from the application of Artificial Neural Network (ANN) and Autoregressive Integrated Moving Average (ARIMA) models are used to forecast a time series. Results: The arithmetic mean, MAPE, overstates the corresponding population parameter, unlike R-MAPE, on a set of error distributions with a statistically significant right skew, as well as outlier values. Conclusions: Our results suggest that R-MAPE represents a suitable alternative measure of forecast accuracy, due to the fact that it provides a valid assessment of forecast accuracy compared to MAPE (AU)


Antecedentes: el Promedio del Error Porcentual Absoluto (MAPE) es probablemente la medida de adecuación de la previsión más ampliamente utilizada. Sin embargo, no cumple el criterio de validez debido a que la distribución de los errores porcentuales absolutos habitualmente presenta una forma asimétrica a la derecha con presencia de valores alejados. En estos casos, el MAPE proporciona una sobreestimación del correspondiente parámetro poblacional. En el presente trabajo se propone un índice alternativo, denominado MAPE Resistente o R-MAPE, y basado en el cálculo del M-estimador de Huber, el cual permite superar la mencionada limitación. Método: se utilizan los resultados derivados de la aplicación de modelos de Red Neuronal Artificial (ANN) y modelos Autorregresivos Integrados de Media Móvil (ARIMA) en la previsión de una serie temporal. Resultados: se puede observar que la media aritmética, el MAPE, realiza una sobreestimación del correspondiente parámetro poblacional, a diferencia del R-MAPE, sobre un conjunto de distribuciones de errores con asimetría a la derecha y presencia de valores alejados. Conclusiones: nuestros resultados ponen de manifiesto que el R-MAPE representa una adecuada alternativa en la medición del ajuste en la previsión, debido a que proporciona una evaluación válida de dicho ajuste en comparación al MAPE (AU)


Assuntos
Humanos , Masculino , Feminino , Psicometria/instrumentação , Psicometria/métodos , Psicometria/tendências , Previsões/métodos , Teorema de Bayes , Previsão do Tempo , Testes Psicológicos/estatística & dados numéricos , Testes Psicológicos/normas , Consumo de Energia/métodos
3.
Psicothema ; 25(4): 500-6, 2013.
Artigo em Inglês | MEDLINE | ID: mdl-24124784

RESUMO

BACKGROUND: The mean absolute percentage error (MAPE) is probably the most widely used goodness-of-fit measure. However, it does not meet the validity criterion due to the fact that the distribution of the absolute percentage errors is usually skewed to the right, with the presence of outlier values. In these cases, MAPE overstates the corresponding population parameter. In this study, we propose an alternative index, called Resistant MAPE or R-MAPE based on the calculation of the Huber M-estimator, which allows overcoming the aforementioned limitation. METHOD: The results derived from the application of Artificial Neural Network (ANN) and Autoregressive Integrated Moving Average (ARIMA) models are used to forecast a time series. RESULTS: The arithmetic mean, MAPE, overstates the corresponding population parameter, unlike R-MAPE, on a set of error distributions with a statistically significant right skew, as well as outlier values. CONCLUSIONS: Our results suggest that R-MAPE represents a suitable alternative measure of forecast accuracy, due to the fact that it provides a valid assessment of forecast accuracy compared to MAPE.


Assuntos
Previsões , Modelos Estatísticos , Redes Neurais de Computação , Algoritmos , Reprodutibilidade dos Testes
4.
Adicciones (Palma de Mallorca) ; 22(4): 293-300, oct.-dic. 2010. tab
Artigo em Inglês | IBECS | ID: ibc-84249

RESUMO

El objetivo del presente estudio es analizar los factores relacionados con el uso de sustancias adictivas en la adolescencia mediante reglas de asociación, herramientas descriptivas incluidas en Data Mining. Para ello se cuenta con una base de datos referidos al consumo de sustancias adictivas en la adolescencia y se utiliza el paquete arules, integrado en el programa de libre distribución R (versión 2.10.1). La muestra está formada por 9.300 estudiantes de edades comprendidas entre los 14 y los 18 años (47,1% chicos y 52,9% chicas) con una edad media de 15,6 años (SE=1,2). Los adolescentes contestaron un cuestionario anónimo que incluía preguntas sobre factores de riesgo personales, familiares y ambientales para el consumo desustancias. Las mejores reglas obtenidas en relación al consumo de sustancias relacionan el consumo de alcohol con la educación paterna percibida y el consumo de los amigos (confianza= 0.8528), el consumo de tabaco, cannabis y cocaína con la actuación paterna percibida y la realización de conductas ilegales (confianzas de 0.8032, 0.8718 y 1.0000, respectivamente)y el uso de éxtasis con el consumo de los iguales (confianza = 1.0000).En general, las reglas de asociación muestran de forma sencilla la relación existente entre ciertas pautas de actuación paterna percibida, la emisión de conductas desviadas de las normas de comportamiento social, el consumo por parte del grupo de iguales y el abuso de drogas, legales e ilegales, en la adolescencia. Se describen las implicaciones de los resultados obtenidos así como la utilidad de esta nueva metodología de análisis (AU)


The aim of this study is to analyse the factors related to the use of addictive substances in adolescence using association rules, descriptive tools included in Data Mining. Thus, we have a database referring to the consumption of addictive substances in adolescence, and use the free distribution program in the R arules package (version 2.10.0). The sample was made up of 9,300 students between the ages of 14 and 18 (47.1% boys and 52.9% girls) with an average age of 15.6 (SE=1.2). The adolescents answered an anonymous questionnaire on personal, family and environmental risk factors related to substance use. The best rules obtained with regard to substance use relate the consumption of alcohol to perceived parenting style and peer consumption (confidence = 0.8528), the use of tobacco (smoking), cannabis and cocaine to perceived parental action and illegal behaviour (confidence= 0.8032, 0.8718 and 1.0000, respectively), and the use of ecstasy to peer consumption (confidence = 1.0000). In general, the association rules show in a simple manner the relationship between certain patterns of perceived parental action, behaviours that deviate from social behavioural norms, peer consumption and the use of different legal and illegal drugs of abuse in adolescence. The implications of the results obtained are described, together with the usefulness of this new methodology of analysis (AU)


Assuntos
Humanos , Masculino , Feminino , Adolescente , Comportamento Aditivo/psicologia , Transtornos Relacionados ao Uso de Substâncias/psicologia , Associação , Comportamento do Adolescente/psicologia , Psicometria/instrumentação , Fumar Maconha/psicologia
5.
Adicciones ; 21(1): 65-80, 2009.
Artigo em Inglês, Espanhol | MEDLINE | ID: mdl-19333526

RESUMO

This paper is aimed mainly at making researchers in the field of drug addictions aware of a methodology of data analysis aimed at knowledge discovery in databases (KDD). KDD is a process consisting of a series of phases, the most characteristic of which is called data mining (DM), whereby different modelling techniques are applied in order to detect patterns and relationships among the data. Common and differentiating factors between the most widely used DM techniques are analysed, mainly from a methodological viewpoint, and their use is exemplified using data related to alcohol consumption in teenagers and its possible relationship with personality variables (N=7030). Although the overall accuracy obtained (% correct predictions) is very similar in the three models analyzed, the Artificial Neural Network (ANN) technique generates the most accurate model (64.1%), followed by Decision Trees (DT) (62.3%) and Naïve Bayes (NB) (59.9%).


Assuntos
Consumo de Bebidas Alcoólicas/psicologia , Redes Neurais de Computação , Adolescente , Algoritmos , Árvores de Decisões , Humanos
6.
Adicciones (Palma de Mallorca) ; 21(1): 65-80, ene.-mar. 2009. ilus, tab
Artigo em Espanhol | IBECS | ID: ibc-61389

RESUMO

El presente trabajo pretende principalmente acercar a los investigadores del campo de las drogodependencias una metodología de análisis de datos orientada al descubrimiento de conocimiento en bases de datos (KDD). El KDD es un proceso que consta de una serie de fases, la más característica de las cuales se denomina Data Mining (DM), en la que se aplican diferentes técnicas de modelado para detectar patrones y relaciones en los datos. Se analizan los factores comunes y diferenciadores de las técnicas DM más ampliamente utilizadas, desde una visión principalmente metodológica, y ejemplificando su uso con datos provenientes del consumo de alcohol en adolescentes y su posible relación con variables de personalidad (N=7030). Aunque la precisión global obtenida (% de predicciones correctas) es muy similar en los tres modelos analizados, las redes neuronales generan el modelo más preciso (64.1%), seguidas de los árboles de decisión (62.3%) y Naive Bayes (59.9%) (AU)


This paper is aimed mainly at making researchers in the field of drug addictions aware of a methodology of data analysis aimed at knowledge discovery in databases (KDD). KDD is a process consisting of a series of phases, the most characteristic of which is called data mining (DM), where by different model ling techniques are applied in order to detect patterns and relationships among the data. Common and differentiating factors between the most widely used DM techniques are analysed, mainly from a methodological viewpoint, and their use is exemplified using data related to alcohol consumption in teenagers and its possible relationship with personality variables (N=7030). Although the overall accuracy obtained (% correct predictions) is very similar in the three models analyzed, the Artificial Neural Network (ANN) technique generates the most accurate model (64.1%), followed by Decision Trees (DT) (62.3%) and Naïve Bayes (NB) (59.9%) (AU)


Assuntos
Humanos , Alcoolismo/epidemiologia , Redes Neurais de Computação , Comportamento do Adolescente , Bases de Dados Estatísticos , Previsões , Árvores de Decisões
SELEÇÃO DE REFERÊNCIAS
DETALHE DA PESQUISA
...