Hybrid Water Demand Forecasting Model Associating Artificial Neural Network with Fourier Series

This paper addressed the problem of water-demand forecasting for real-time operation of water supply systems. The present study was conducted to identify the best fit model using hourly consumption data from the water supply system of Araraquara, Sa approximate to o Paulo, Brazil. Artificial neural networks (ANNs) were used in view of their enhanced capability to match or even improve on the regression model forecasts. The ANNs used were the multilayer perceptron with the back-propagation algorithm (MLP-BP), the dynamic neural network (DAN2), and two hybrid ANNs. The hybrid models used the error produced by the Fourier series forecasting as input to the MLP-BP and DAN2, called ANN-H and DAN2-H, respectively. The tested inputs for the neural network were selected literature and correlation analysis. The results from the hybrid models were promising, DAN2 performing better than the tested MLP-BP models. DAN2-H, identified as the best model, produced a mean absolute error (MAE) of 3.3 L/s and 2.8 L/s for training and test set, respectively, for the prediction of the next hour, which represented about 12% of the average consumption. The best forecasting model for the next 24 hours was again DAN2-H, which outperformed other compared models, and produced a MAE of 3.1 L/s and 3.0 L/s for training and test set respectively, which represented about 12% of average consumption. DOI: 10.1061/(ASCE)WR.1943-5452.0000177. (C) 2012 American Society of Civil Engineers.

Independent predictors and a prognostic model for surgical outcome in refractory frontal lobe epilepsy

Purpose: Refractory frontal lobe epilepsy (FLE) remains one of the most challenging surgically remediable epilepsy syndromes. Nevertheless, definition of independent predictors and predictive models of postsurgical seizure outcome remains poorly explored in FLE. Methods: We retrospectively analyzed data from 70 consecutive patients with refractory FLE submitted to surgical treatment at our center from July 1994 to December 2006. Univariate results were submitted to logistic regression models and Cox proportional hazards regression to identify isolated risk factors for poor surgical results and to construct predictive models for surgical outcome in FLE. Results: From 70 patients submitted to surgery, 45 patients (64%) had favorable outcome and 37 (47%) became seizure free. Isolated risk factors for poor surgical outcome are expressed in hazard ratio (H.R.) and were time of epilepsy (H.R.=4.2; 95% C.I.=.1.5-11.7; p=0.006), ictal EEG recruiting rhythm (H.R. = 2.9; 95% C.I. = 1.1-7.7; p=0.033); normal MRI (H.R. = 4.8; 95% C.I. = 1.4-16.6; p = 0.012), and MRI with lesion involving eloquent cortex (H.R. = 3.8; 95% C.I. = 1.2-12.0; p = 0.021). Based on these variables and using a logistic regression model we constructed a model that correctly predicted long-term surgical outcome in up to 80% of patients. Conclusion: Among independent risk factors for postsurgical seizure outcome, epilepsy duration is a potentially modifiable factor that could impact surgical outcome in FLE. Early diagnosis, presence of an MRI lesion not involving eloquent cortex, and ictal EEG without recruited rhythm independently predicted favorable outcome in this series. (C) 2011 Elsevier B.V. All rights reserved.

Why Are Women With Cervical Cancer Not Being Diagnosed in Preinvasive Phase? An Analysis of Risk Factors Using a Hierarchical Model

Objective: To assess the risk factors for delayed diagnosis of uterine cervical lesions. Materials and Methods: This is a case-control study that recruited 178 women at 2 Brazilian hospitals. The cases (n = 74) were composed of women with a late diagnosis of a lesion in the uterine cervix (invasive carcinoma in any stage). The controls (n = 104) were composed of women with cervical lesions diagnosed early on (low-or high-grade intraepithelial lesions). The analysis was performed by means of logistic regression model using a hierarchical model. The socioeconomic and demographic variables were included at level I (distal). Level II (intermediate) included the personal and family antecedents and knowledge about the Papanicolaou test and human papillomavirus. Level III (proximal) encompassed the variables relating to individuals' care for their own health, gynecologic symptoms, and variables relating to access to the health care system. Results: The risk factors for late diagnosis of uterine cervical lesions were age older than 40 years (odds ratio [OR] = 10.4; 95% confidence interval [CI], 2.3-48.4), not knowing the difference between the Papanicolaou test and gynecological pelvic examinations (OR, = 2.5; 95% CI, 1.3-4.9), not thinking that the Papanicolaou test was important (odds ratio [OR], 4.2; 95% CI, 1.3-13.4), and abnormal vaginal bleeding (OR, 15.0; 95% CI, 6.5-35.0). Previous treatment for sexually transmissible disease was a protective factor (OR, 0.3; 95% CI, 0.1-0.8) for delayed diagnosis. Conclusions: Deficiencies in cervical cancer prevention programs in developing countries are not simply a matter of better provision and coverage of Papanicolaou tests. The misconception about the Papanicolaou test is a serious educational problem, as demonstrated by the present study.

A Multilevel Model with Time Series Components for the Analysis of Tribal Art Prices

In the present work we perform an econometric analysis of the Tribal art market. To this aim, we use a unique and original database that includes information on Tribal art market auctions worldwide from 1998 to 2011. In Literature, art prices are modelled through the hedonic regression model, a classic fixed-effect model. The main drawback of the hedonic approach is the large number of parameters, since, in general, art data include many categorical variables. In this work, we propose a multilevel model for the analysis of Tribal art prices that takes into account the influence of time on artwork prices. In fact, it is natural to assume that time exerts an influence over the price dynamics in various ways. Nevertheless, since the set of objects change at every auction date, we do not have repeated measurements of the same items over time. Hence, the dataset does not constitute a proper panel; rather, it has a two-level structure in that items, level-1 units, are grouped in time points, level-2 units. The main theoretical contribution is the extension of classical multilevel models to cope with the case described above. In particular, we introduce a model with time dependent random effects at the second level. We propose a novel specification of the model, derive the maximum likelihood estimators and implement them through the E-M algorithm. We test the finite sample properties of the estimators and the validity of the own-written R-code by means of a simulation study. Finally, we show that the new model improves considerably the fit of the Tribal art data with respect to both the hedonic regression model and the classic multilevel model.

A prediction model for assessing residential radon concentration in Switzerland

Indoor radon is regularly measured in Switzerland. However, a nationwide model to predict residential radon levels has not been developed. The aim of this study was to develop a prediction model to assess indoor radon concentrations in Switzerland. The model was based on 44,631 measurements from the nationwide Swiss radon database collected between 1994 and 2004. Of these, 80% randomly selected measurements were used for model development and the remaining 20% for an independent model validation. A multivariable log-linear regression model was fitted and relevant predictors selected according to evidence from the literature, the adjusted R², the Akaike's information criterion (AIC), and the Bayesian information criterion (BIC). The prediction model was evaluated by calculating Spearman rank correlation between measured and predicted values. Additionally, the predicted values were categorised into three categories (50th, 50th-90th and 90th percentile) and compared with measured categories using a weighted Kappa statistic. The most relevant predictors for indoor radon levels were tectonic units and year of construction of the building, followed by soil texture, degree of urbanisation, floor of the building where the measurement was taken and housing type (P-values <0.001 for all). Mean predicted radon values (geometric mean) were 66 Bq/m³ (interquartile range 40-111 Bq/m³) in the lowest exposure category, 126 Bq/m³ (69-215 Bq/m³) in the medium category, and 219 Bq/m³ (108-427 Bq/m³) in the highest category. Spearman correlation between predictions and measurements was 0.45 (95%-CI: 0.44; 0.46) for the development dataset and 0.44 (95%-CI: 0.42; 0.46) for the validation dataset. Kappa coefficients were 0.31 for the development and 0.30 for the validation dataset, respectively. The model explained 20% overall variability (adjusted R²). In conclusion, this residential radon prediction model, based on a large number of measurements, was demonstrated to be robust through validation with an independent dataset. The model is appropriate for predicting radon level exposure of the Swiss population in epidemiological research. Nevertheless, some exposure misclassification and regression to the mean is unavoidable and should be taken into account in future applications of the model.

Model Evaluation Based on the Distribution of Estimated Absolute Prediction Error

The construction of a reliable, practically useful prediction rule for future response is heavily dependent on the "adequacy" of the fitted regression model. In this article, we consider the absolute prediction error, the expected value of the absolute difference between the future and predicted responses, as the model evaluation criterion. This prediction error is easier to interpret than the average squared error and is equivalent to the mis-classification error for the binary outcome. We show that the distributions of the apparent error and its cross-validation counterparts are approximately normal even under a misspecified fitted model. When the prediction rule is "unsmooth", the variance of the above normal distribution can be estimated well via a perturbation-resampling method. We also show how to approximate the distribution of the difference of the estimated prediction errors from two competing models. With two real examples, we demonstrate that the resulting interval estimates for prediction errors provide much more information about model adequacy than the point estimates alone.

A Diagnostic Test for the Mixing Distribution in a Generalised Linear Mixed Model

We introduce a diagnostic test for the mixing distribution in a generalised linear mixed model. The test is based on the difference between the marginal maximum likelihood and conditional maximum likelihood estimates of a subset of the fixed effects in the model. We derive the asymptotic variance of this difference, and propose a test statistic that has a limiting chi-square distribution under the null hypothesis that the mixing distribution is correctly specified. For the important special case of the logistic regression model with random intercepts, we evaluate via simulation the power of the test in finite samples under several alternative distributional forms for the mixing distribution. We illustrate the method by applying it to data from a clinical trial investigating the effects of hormonal contraceptives in women.

Checking Assumptions in Latent Class Regression Models via a Markov Chain Monte Carlo Estimation Approach: An Application to Depression and Socio-Economic Status

Latent class regression models are useful tools for assessing associations between covariates and latent variables. However, evaluation of key model assumptions cannot be performed using methods from standard regression models due to the unobserved nature of latent outcome variables. This paper presents graphical diagnostic tools to evaluate whether or not latent class regression models adhere to standard assumptions of the model: conditional independence and non-differential measurement. An integral part of these methods is the use of a Markov Chain Monte Carlo estimation procedure. Unlike standard maximum likelihood implementations for latent class regression model estimation, the MCMC approach allows us to calculate posterior distributions and point estimates of any functions of parameters. It is this convenience that allows us to provide the diagnostic methods that we introduce. As a motivating example we present an analysis focusing on the association between depression and socioeconomic status, using data from the Epidemiologic Catchment Area study. We consider a latent class regression analysis investigating the association between depression and socioeconomic status measures, where the latent variable depression is regressed on education and income indicators, in addition to age, gender, and marital status variables. While the fitted latent class regression model yields interesting results, the model parameters are found to be invalid due to the violation of model assumptions. The violation of these assumptions is clearly identified by the presented diagnostic plots. These methods can be applied to standard latent class and latent class regression models, and the general principle can be extended to evaluate model assumptions in other types of models.

Towards greater accuracy in individual-tree mortality regression

Background mortality is an essential component of any forest growth and yield model. Forecasts of mortality contribute largely to the variability and accuracy of model predictions at the tree, stand and forest level. In the present study, I implement and evaluate state-of-the-art techniques to increase the accuracy of individual tree mortality models, similar to those used in many of the current variants of the Forest Vegetation Simulator, using data from North Idaho and Montana. The first technique addresses methods to correct for bias induced by measurement error typically present in competition variables. The second implements survival regression and evaluates its performance against the traditional logistic regression approach. I selected the regression calibration (RC) algorithm as a good candidate for addressing the measurement error problem. Two logistic regression models for each species were fitted, one ignoring the measurement error, which is the “naïve” approach, and the other applying RC. The models fitted with RC outperformed the naïve models in terms of discrimination when the competition variable was found to be statistically significant. The effect of RC was more obvious where measurement error variance was large and for more shade-intolerant species. The process of model fitting and variable selection revealed that past emphasis on DBH as a predictor variable for mortality, while producing models with strong metrics of fit, may make models less generalizable. The evaluation of the error variance estimator developed by Stage and Wykoff (1998), and core to the implementation of RC, in different spatial patterns and diameter distributions, revealed that the Stage and Wykoff estimate notably overestimated the true variance in all simulated stands, but those that are clustered. Results show a systematic bias even when all the assumptions made by the authors are guaranteed. I argue that this is the result of the Poisson-based estimate ignoring the overlapping area of potential plots around a tree. Effects, especially in the application phase, of the variance estimate justify suggested future efforts of improving the accuracy of the variance estimate. The second technique implemented and evaluated is a survival regression model that accounts for the time dependent nature of variables, such as diameter and competition variables, and the interval-censored nature of data collected from remeasured plots. The performance of the model is compared with the traditional logistic regression model as a tool to predict individual tree mortality. Validation of both approaches shows that the survival regression approach discriminates better between dead and alive trees for all species. In conclusion, I showed that the proposed techniques do increase the accuracy of individual tree mortality models, and are a promising first step towards the next generation of background mortality models. I have also identified the next steps to undertake in order to advance mortality models further.

Biodiversity effects on nitrate concentrations in soil solution: a Bayesian model

Ecosystems are faced with high rates of species loss which has consequences for their functions and services. To assess the effects of plant species diversity on the nitrogen (N) cycle, we developed a model for monthly mean nitrate (NO3-N) concentrations in soil solution in 0-30 cm mineral soil depth using plant species and functional group richness and functional composition as drivers and assessing the effects of conversion of arable land to grassland, spatially heterogeneous soil properties, and climate. We used monthly mean NO3-N concentrations from 62 plots of a grassland plant diversity experiment from 2003 to 2006. Plant species richness (1-60) and functional group composition (1-4 functional groups: legumes, grasses, non-leguminous tall herbs, non-leguminous small herbs) were manipulated in a factorial design. Plant community composition, time since conversion from arable land to grassland, soil texture, and climate data (precipitation, soil moisture, air and soil temperature) were used to develop one general Bayesian multiple regression model for the 62 plots to allow an in-depth evaluation using the experimental design. The model simulated NO3-N concentrations with an overall Bayesian coefficient of determination of 0.48. The temporal course of NO3-N concentrations was simulated differently well for the individual plots with a maximum plot-specific Nash-Sutcliffe Efficiency of 0.57. The model shows that NO3-N concentrations decrease with species richness, but this relation reverses if more than approx. 25 % of legume species are included in the mixture. Presence of legumes increases and presence of grasses decreases NO3-N concentrations compared to mixtures containing only small and tall herbs. Altogether, our model shows that there is a strong influence of plant community composition on NO3-N concentrations.

Independent at heart: persistent association of altitude with ischaemic heart disease mortality after consideration of climate, topography and built environment.

BACKGROUND Living at higher altitude was dose-dependently associated with lower risk of ischaemic heart disease (IHD). Higher altitudes have different climatic, topographic and built environment properties than lowland regions. It is unclear whether these environmental factors mediate/confound the association between altitude and IHD. We examined how much of the altitude-IHD association is explained by variations in exposure at place of residence to sunshine, temperature, precipitation, aspect, slope and distance to main road. METHODS We included 4.2 million individuals aged 40-84 at baseline living in Switzerland at altitudes 195-2971 m above sea level (ie, full range of residence), providing 77 127 IHD deaths. Mortality data 2000-2008, sociodemographic/economic information and coordinates of residence were obtained from the Swiss National Cohort, a longitudinal, census-based record linkage study. Environment information was modelled to residence level using Weibull regression models. RESULTS In the model not adjusted for other environmental factors, IHD mortality linearly decreased with increasing altitude resulting in a lower risk (HR, 95% CI 0.67, 0.60 to 0.74) for those living >1500 m (vs<600 m). This association remained after adjustment for all other environmental factors 0.74 (0.66 to 0.82). CONCLUSIONS The benefit of living at higher altitude was only partially confounded by variations in climate, topography and built environment. Rather, physical environment factors appear to have an independent effect and may impact on cardiovascular health in a cumulative way. Inclusion of additional modifiable factors as well as individual information on traditional IHD risk factors in our combined environmental model could help to identify strategies for the reduction of inequalities in IHD mortality.

Efficiency Bounds for Estimating Linear Functionals of Nonparametric Regression Models with Endogenous Regressors

Consider a nonparametric regression model Y=mu*(X) + e, where the explanatory variables X are endogenous and e satisfies the conditional moment restriction E[e|W]=0 w.p.1 for instrumental variables W. It is well known that in these models the structural parameter mu* is 'ill-posed' in the sense that the function mapping the data to mu* is not continuous. In this paper, we derive the efficiency bounds for estimating linear functionals E[p(X)mu*(X)] and int_{supp(X)}p(x)mu*(x)dx, where p is a known weight function and supp(X) the support of X, without assuming mu* to be well-posed or even identified.

Selection bias and the cross-validation of regression models for prediction

Strategies are compared for the development of a linear regression model with stochastic (multivariate normal) regressor variables and the subsequent assessment of its predictive ability. Bias and mean squared error of four estimators of predictive performance are evaluated in simulated samples of 32 population correlation matrices. Models including all of the available predictors are compared with those obtained using selected subsets. The subset selection procedures investigated include two stopping rules, C$\sb{\rm p}$ and S$\sb{\rm p}$, each combined with an 'all possible subsets' or 'forward selection' of variables. The estimators of performance utilized include parametric (MSEP$\sb{\rm m}$) and non-parametric (PRESS) assessments in the entire sample, and two data splitting estimates restricted to a random or balanced (Snee's DUPLEX) 'validation' half sample. The simulations were performed as a designed experiment, with population correlation matrices representing a broad range of data structures.^ The techniques examined for subset selection do not generally result in improved predictions relative to the full model. Approaches using 'forward selection' result in slightly smaller prediction errors and less biased estimators of predictive accuracy than 'all possible subsets' approaches but no differences are detected between the performances of C$\sb{\rm p}$ and S$\sb{\rm p}$. In every case, prediction errors of models obtained by subset selection in either of the half splits exceed those obtained using all predictors and the entire sample.^ Only the random split estimator is conditionally (on $\\beta$) unbiased, however MSEP$\sb{\rm m}$ is unbiased on average and PRESS is nearly so in unselected (fixed form) models. When subset selection techniques are used, MSEP$\sb{\rm m}$ and PRESS always underestimate prediction errors, by as much as 27 percent (on average) in small samples. Despite their bias, the mean squared errors (MSE) of these estimators are at least 30 percent less than that of the unbiased random split estimator. The DUPLEX split estimator suffers from large MSE as well as bias, and seems of little value within the context of stochastic regressor variables.^ To maximize predictive accuracy while retaining a reliable estimate of that accuracy, it is recommended that the entire sample be used for model development, and a leave-one-out statistic (e.g. PRESS) be used for assessment. ^

REGRESSION ANALYSIS FOR STANDARDIZED MORTALITY RATIOS WITH APPLICATION TO OCCUPATIONAL FOLLOW-UP STUDIES

Traditional comparison of standardized mortality ratios (SMRs) can be misleading if the age-specific mortality ratios are not homogeneous. For this reason, a regression model has been developed which incorporates the mortality ratio as a function of age. This model is then applied to mortality data from an occupational cohort study. The nature of the occupational data necessitates the investigation of mortality ratios which increase with age. These occupational data are used primarily to illustrate and develop the statistical methodology.^ The age-specific mortality ratio (MR) for the covariates of interest can be written as MR(,ij...m) = ((mu)(,ij...m)/(theta)(,ij...m)) = r(.)exp (Z('')(,ij...m)(beta)) where (mu)(,ij...m) and (theta)(,ij...m) denote the force of mortality in the study and chosen standard populations in the ij...m('th) stratum, respectively, r is the intercept, Z(,ij...m) is the vector of covariables associated with the i('th) age interval, and (beta) is a vector of regression coefficients associated with these covariables. A Newton-Raphson iterative procedure has been used for determining the maximum likelihood estimates of the regression coefficients.^ This model provides a statistical method for a logical and easily interpretable explanation of an occupational cohort mortality experience. Since it gives a reasonable fit to the mortality data, it can also be concluded that the model is fairly realistic. The traditional statistical method for the analysis of occupational cohort mortality data is to present a summary index such as the SMR under the assumption of constant (homogeneous) age-specific mortality ratios. Since the mortality ratios for occupational groups usually increase with age, the homogeneity assumption of the age-specific mortality ratios is often untenable. The traditional method of comparing SMRs under the homogeneity assumption is a special case of this model, without age as a covariate.^ This model also provides a statistical technique to evaluate the relative risk between two SMRs or a dose-response relationship among several SMRs. The model presented has application in the medical, demographic and epidemiologic areas. The methods developed in this thesis are suitable for future analyses of mortality or morbidity data when the age-specific mortality/morbidity experience is a function of age or when there is an interaction effect between confounding variables needs to be evaluated. ^

Slope estimation in a parametric measurement error model: A simulation study

In regression analysis, covariate measurement error occurs in many applications. The error-prone covariates are often referred to as latent variables. In this proposed study, we extended the study of Chan et al. (2008) on recovering latent slope in a simple regression model to that in a multiple regression model. We presented an approach that applied the Monte Carlo method in the Bayesian framework to the parametric regression model with the measurement error in an explanatory variable. The proposed estimator applied the conditional expectation of latent slope given the observed outcome and surrogate variables in the multiple regression models. A simulation study was presented showing that the method produces estimator that is efficient in the multiple regression model, especially when the measurement error variance of surrogate variable is large.^