Bayesian model selection methodology for road safety

Road accidents are a very relevant issue in many countries and macroeconomic models are very frequently applied by academia and administrations to reduce their frequency and consequences. The selection of explanatory variables and response transformation parameter within the Bayesian framework for the selection of the set of explanatory variables a TIM and 3IM (two input and three input models) procedures are proposed. The procedure also uses the DIC and pseudo -R2 goodness of fit criteria. The model to which the methodology is applied is a dynamic regression model with Box-Cox transformation (BCT) for the explanatory variables and autorgressive (AR) structure for the response. The initial set of 22 explanatory variables are identified. The effects of these factors on the fatal accident frequency in Spain, during 2000-2012, are estimated. The dependent variable is constructed considering the stochastic trend component.

Geografía del carbono en alta resolución en bosque tropical amazónico del Ecuador mediante sensores aerotransportados

La estimación de la biomasa de la vegetación terrestre en bosque tropical no sólo es un área de investigación en rápida expansión, sino también es un tema de gran interés para reducir las emisiones de carbono asociadas a la deforestación y la degradación forestal (REDD+). Las estimaciones de densidad de carbono sobre el suelo (ACD) en base a inventarios de campo y datos provenientes de sensores aerotransportados, en especial con sensores LiDAR, han conducido a un progreso sustancial en el cartografiado a gran escala de las reservas de carbono forestal. Sin embargo, estos mapas de carbono tienen incertidumbres considerables, asociadas generalmente al proceso de calibración del modelo de regresión utilizado para producir los mapas. En esta tesis se establece una metodología para la calibración y validación de un modelo general de estimación de ACD usando LiDAR en un sector del Parque Nacional Yasuní en Ecuador. En el proceso de calibración del modelo se considera el tamaño y la ubicación de las parcelas, la influencia de la topografía y la distribución espacial de la biomasa. Para el análisis de los datos se utilizan técnicas geoestadísticas en combinación con variables geomorfométricas derivadas de datos LiDAR, y se propone un esquema de muestreo estratificado por posiciones topográficas (valle, ladera y cima). La validación del modelo general para toda la zona de estudio presentó valores de RMSE = 5.81 Mg C ha-1, R2 = 0.94 y sesgo = 0.59, mientras que, al considerar las posiciones topográficas, el modelo presentó valores de RMSE = 1.67 Mg C ha-1, R2 = 0.98 y sesgo = 0.23 para el valle; RMSE = 3.13 Mg C ha-1, R2 = 0.98 y sesgo = - 0.34 para la ladera; y RMSE = 2.33 Mg C ha-1, R2 = 0.97 y sesgo = 0.74 para la cima. Los resultados obtenidos demuestran que la metodología de muestreo estratificado por posiciones topográficas propuesto, permite calibrar de manera efectiva el modelo general con las estimaciones de ACD en campo, logrando reducir el RMSE y el sesgo. Los resultados muestran el potencial de los datos LiDAR para caracterizar la estructura vertical de la vegetación en un bosque altamente diverso, permitiendo realizar estimaciones precisas de ACD, y conocer patrones espaciales continuos de la distribución de la biomasa aérea y del contenido de carbono en la zona de estudio. ABSTRACT Estimating biomass of terrestrial vegetation in tropical forest is not only a rapidly expanding research area, but also a subject of tremendous interest for reducing carbon emissions associated with deforestation and forest degradation (REDD+). The aboveground carbon density estimates (ACD) based on field inventories and airborne sensors, especially LiDAR sensors have led to a substantial progress in large-scale mapping of forest carbon stocks. However, these carbon maps have considerable uncertainties generally associated with the calibration of the regression model used to produce these maps. This thesis establishes a methodology for calibrating and validating a general ACD estimation model using LiDAR in Ecuador´s Yasuní National Park. The size and location of the plots are considered in the model calibration phase as well as the influence of topography and spatial distribution of biomass. Geostatistical analysis techniques are used in combination with geomorphometrics variables derived from LiDAR data, and then a stratified sampling scheme considering topographic positions (valley, slope and ridge) is proposed. The validation of the general model for the study area showed values of RMSE = 5.81 Mg C ha-1, R2 = 0.94 and bias = 0.59, while considering the topographical positions, the model showed values of RMSE = 1.67 Mg C ha-1, R2 = 0.98 and bias = 0.23 for the valley; RMSE = 3.13 Mg C ha-1, R2 = 0.98 and bias = - 0.34 for the slope; and RMSE = 2.33 Mg C ha-1, R2 = 0.97 and bias = 0.74 for the ridge. The results show that the stratified sampling methodology taking into account topographic positions, effectively calibrates the general model with field estimates of ACD, reducing RMSE and bias. The results show the potential of LiDAR data to characterize the vertical structure of vegetation in a highly diverse forest, allowing accurate estimates of ACD, and knowing continuous spatial patterns of biomass distribution and carbon stocks in the study area.

WGTTEST: Stata module to test the impact of sampling weights in regression analysis

wgttest performs a test proposed by DuMouchel and Duncan (1983) to evaluate whether the weighted and unweighted estimates of a regression model are significantly different.

Multi-level zero-inflated Poisson regression modelling of correlated count data with excess zeros

Count data with excess zeros relative to a Poisson distribution are common in many biomedical applications. A popular approach to the analysis of such data is to use a zero-inflated Poisson (ZIP) regression model. Often, because of the hierarchical Study design or the data collection procedure, zero-inflation and lack of independence may occur simultaneously, which tender the standard ZIP model inadequate. To account for the preponderance of zero counts and the inherent correlation of observations, a class of multi-level ZIP regression model with random effects is presented. Model fitting is facilitated using an expectation-maximization algorithm, whereas variance components are estimated via residual maximum likelihood estimating equations. A score test for zero-inflation is also presented. The multi-level ZIP model is then generalized to cope with a more complex correlation structure. Application to the analysis of correlated count data from a longitudinal infant feeding study illustrates the usefulness of the approach.

Optimal crossover designs for logistic regression models in pharmacodynamics

Pharmacodynamics (PD) is the study of the biochemical and physiological effects of drugs. The construction of optimal designs for dose-ranging trials with multiple periods is considered in this paper, where the outcome of the trial (the effect of the drug) is considered to be a binary response: the success or failure of a drug to bring about a particular change in the subject after a given amount of time. The carryover effect of each dose from one period to the next is assumed to be proportional to the direct effect. It is shown for a logistic regression model that the efficiency of optimal parallel (single-period) or crossover (two-period) design is substantially greater than a balanced design. The optimal designs are also shown to be robust to misspecification of the value of the parameters. Finally, the parallel and crossover designs are combined to provide the experimenter with greater flexibility.

Guiding local regression using visualisation

Solving many scientific problems requires effective regression and/or classification models for large high-dimensional datasets. Experts from these problem domains (e.g. biologists, chemists, financial analysts) have insights into the domain which can be helpful in developing powerful models but they need a modelling framework that helps them to use these insights. Data visualisation is an effective technique for presenting data and requiring feedback from the experts. A single global regression model can rarely capture the full behavioural variability of a huge multi-dimensional dataset. Instead, local regression models, each focused on a separate area of input space, often work better since the behaviour of different areas may vary. Classical local models such as Mixture of Experts segment the input space automatically, which is not always effective and it also lacks involvement of the domain experts to guide a meaningful segmentation of the input space. In this paper we addresses this issue by allowing domain experts to interactively segment the input space using data visualisation. The segmentation output obtained is then further used to develop effective local regression models.

Guiding local regression using visualisation

Solving many scientific problems requires effective regression and/or classification models for large high-dimensional datasets. Experts from these problem domains (e.g. biologists, chemists, financial analysts) have insights into the domain which can be helpful in developing powerful models but they need a modelling framework that helps them to use these insights. Data visualisation is an effective technique for presenting data and requiring feedback from the experts. A single global regression model can rarely capture the full behavioural variability of a huge multi-dimensional dataset. Instead, local regression models, each focused on a separate area of input space, often work better since the behaviour of different areas may vary. Classical local models such as Mixture of Experts segment the input space automatically, which is not always effective and it also lacks involvement of the domain experts to guide a meaningful segmentation of the input space. In this paper we addresses this issue by allowing domain experts to interactively segment the input space using data visualisation. The segmentation output obtained is then further used to develop effective local regression models.

Examining links between cognitive markers, movement initiation and change, and pedestrian safety in older adults

Objective: The purpose of this study was to determine the extent to which mobility indices (such as walking speed and postural sway), motor initiation, and cognitive function, specifically executive functions, including spatial planning, visual attention, and within participant variability, differentially predicted collisions in the near and far sides of the road with increasing age. Methods: Adults aged over 45 years participated in cognitive tests measuring executive function and visual attention (using Useful Field of View; UFoV®), mobility assessments (walking speed, sit-to-stand, self-reported mobility, and postural sway assessed using motion capture cameras), and gave road crossing choices in a two-way filmed real traffic pedestrian simulation. Results: A stepwise regression model of walking speed, start-up delay variability, and processing speed) explained 49.4% of the variance in near-side crossing errors. Walking speed, start-up delay measures (average & variability), and spatial planning explained 54.8% of the variance in far-side unsafe crossing errors. Start-up delay was predicted by walking speed only (explained 30.5%). Conclusion: Walking speed and start-up delay measures were consistent predictors of unsafe crossing behaviours. Cognitive measures, however, differentially predicted near-side errors (processing speed), and far-side errors (spatial planning). These findings offer potential contributions for identifying and rehabilitating at-risk older pedestrians.

Local-Median Method of Forecasting for Regression Time Series under Outliers

2002 Mathematics Subject Classification: 62M20, 62-07, 62J05, 62P20.

Application of the D-Fullness Technique for Breakdown Point Study of the Trimmed Likelihood Estimator to a Generalized Logistic Model

2000 Mathematics Subject Classification: 62J12, 62F35

Gázolaj-jövedékiadó verseny az Európai Unióban (Fiscal competition on the market for diesel fuel in the European Union)

A tanulmány azt a kérdést vizsgálja, hogy versenyeznek-e az európai kormányok gázolajra vonatkozó jövedékiadó-kulcsaikkal a nagyobb adóbevételekért, és ha igen, befolyásolja-e az országok mérete kormányaik adóztatási stratégiáját. Az üzemanyagturizmussal szembesülő kormányok adókivetési magatartását egy kétországos adóverseny modellel jelezzük előre, amelyben a standard modellektől eltérően a fogyasztók kereslete árrugalmas. Megmutatjuk, hogy ha a kereslet nem teljesen rugalmatlan, mint Nielsen [2001], illetve Kanbur-Keen [1993] modelljeiben, akkor a nagy ország kormányának egyensúlyi viselkedése nemcsak abban különbözik a kicsiétől, hogy nagyobb adót állapít meg, hanem abban is, hogy válaszfüggvénye meredekebb. Az aszimmetrikus adóverseny általunk használt modelljét a dízelüzemanyagoknak 16 európai ország 1978 és 2005 közötti jövedékiadó-adatain vizsgáljuk. Az 1995 és 2005 közötti időszakra vonatkozó becslési eredményeink megerősítik, hogy az európai országok szomszédaik adókulcs-változtatásának hatására változtattak saját adókulcsaikon, és hogy a területileg/gazdaságilag kisebb országok kisebb intenzitással reagáltak szomszédaik adóváltoztatásra, mint a nagyobbak. Tanulmányunk ezzel magyarázatot nyújt arra is, hogy miért erősödött fel a tagállamok jövedéki adókulcsainak méret szerinti differenciálódása az elmúlt bő tíz évben, valamint hogy miért nem sikerült az Európai Uniónak a minimumadószintre vonatkozó előírásával előbbre lépnie az egységes adóztatás megvalósításában. / === / The paper assesses spatial competition in diesel taxation among European governments. By adding an extension to the standard model, it is shown that asymmetric competition – small countries undercutting large – implies that small countries respond less strongly to tax changes by their neighbours than large countries do. An estimate is then made of the fiscal reaction functions for national governments, employing a first-difference regression model with a weighting scheme constructed from road-traffic density data at national borders. Data from 16 countries (EU-15 minus Greece plus Norway and Switzerland) between 1978 and 2005 provides evidence that European governments set their diesel tax interdependently, and moreover, that small European countries tend to react less strongly to changes in their competitors' tax rate than large countries do.

Regression games

The solution of a TU cooperative game can be a distribution of the value of the grand coalition, i.e. it can be a distribution of the payo (utility) all the players together achieve. In a regression model, the evaluation of the explanatory variables can be a distribution of the overall t, i.e. the t of the model every regressor variable is involved. Furthermore, we can take regression models as TU cooperative games where the explanatory (regressor) variables are the players. In this paper we introduce the class of regression games, characterize it and apply the Shapley value to evaluating the explanatory variables in regression models. In order to support our approach we consider Young (1985)'s axiomatization of the Shapley value, and conclude that the Shapley value is a reasonable tool to evaluate the explanatory variables of regression models.

The use of Poisson regression in the sociological study of suicide

This paper explains how Poisson regression can be used in studies in which the dependent variable describes the number of occurrences of some rare event such as suicide. After pointing out why ordinary linear regression is inappropriate for treating dependent variables of this sort, we go on to present the basic Poisson regression model and show how it fits in the broad class of generalized linear models. Then we turn to discussing a major problem of Poisson regression known as overdispersion and suggest possible solutions, including the correction of standard errors and negative binomial regression. The paper ends with a detailed empirical example, drawn from our own research on suicide.

Spatiotemporal variation of the abundance of calcareous green macroalgae in the Florida Keys: A study of synchrony within a macroalgal functional-form group

The abundance of calcareous green algae was recorded quarterly at 28 sites within the Florida Keys National Marine Sanctuary (FKNMS) for a period of 7 years as part of a sea grass monitoring program. To evaluate the validity of using the functional-form group approach, we designed a sampling method that included the functional-form group and the component genera. This strategy enabled us to analyze the spatiotemporal patterns in the abundance of calcareous green algae as a group and to describe synchronous behavior among its genera through the application of a nonlinear regression model to both categories of data. Spatial analyses revealed that, in general, all genera displayed long-term trends of increasing abundance at most sites; however, at some sites the long-term trends for genera opposed one another. Strong synchrony in the timing of seasonal changes was found among all genera, possibly reflecting similar reproductive and seasonal growth pattern, but the variability in the magnitude of seasonal changes was very high among genera and sites. No spatial patterns were found in long-term or seasonal changes; the only significant relation detected was for slope, with sites closer to land showing higher values, suggesting that some factors associated with land proximity are affecting this increase. We conclude that the abundances of genera behaved differently from the functional-form group, indicating that the use of the functionalform group approach may be unsuitable to detect changes in sea grass community structure in the FKNMS at the existing temporal and spatial scale of the monitoring program.

Variables affecting the success of older applicants for employment in a public organization: An age -cued cognitive model of administrative decision-making

The dissertation takes a multivariate approach to answer the question of how applicant age, after controlling for other variables, affects employment success in a public organization. In addition to applicant age, there are five other categories of variables examined: organization/applicant variables describing the relationship of the applicant to the organization; organization/position variables describing the target position as it relates to the organization; episodic variables such as applicant age relative to the ages of competing applicants; economic variables relating to the salary needs of older applicants; and cognitive variables that may affect the decision maker's evaluation of the applicant. ^ An exploratory phase of research employs archival data from approximately 500 decisions made in the past three years to hire or promote applicants for positions in one public health administration organization. A logit regression model is employed to examine the probability that the variables modify the effect of applicant age on employment success. A confirmatory phase of the dissertation is a controlled experiment in which hiring decision makers from the same public organization perform a simulated hiring decision exercise to evaluate hypothetical applicants of similar qualifications but of different ages. The responses of the decision makers to a series of bipolar adjective scales add support to the cognitive component of the theoretical model of the hiring decision. A final section contains information gathered from interviews with key informants. ^ Applicant age has tended to have a curvilinear relationship with employment success. For some positions, the mean age of the applicants most likely to succeed varies with the values of the five groups of moderating variables. The research contributes not only to the practice of public personnel administration, but is useful in examining larger public policy issues associated with an aging workforce. ^