Home About us Contact
|
Home About us Contact | ||
|
|
||
Misclassification Rates (misclassification + rate)
Selected AbstractsArtificial neural networks as statistical tools in epidemiological studies: analysis of risk factors for early infant wheezePAEDIATRIC & PERINATAL EPIDEMIOLOGY, Issue 6 2004Andrea Sherriff Summary Artificial neural networks (ANNs) are being used increasingly for the prediction of clinical outcomes and classification of disease phenotypes. A lack of understanding of the statistical principles underlying ANNs has led to widespread misuse of these tools in the biomedical arena. In this paper, the authors compare the performance of ANNs with that of conventional linear logistic regression models in an epidemiological study of infant wheeze. Data on the putative risk factors for infant wheeze have been obtained from a sample of 7318 infants taking part in the Avon Longitudinal Study of Parents and Children (ALSPAC). The data were analysed using logistic regression models and ANNs, and performance based on misclassification rates of a validation data set were compared. Misclassification rates in the training data set decreased as the complexity of the ANN increased: h = 0: 17.9%; h = 2: 16.2%; h = 5: 14.9%, and h = 10: 9.2%. However, the more complex models did not generalise well to new data sets drawn from the same population: validation data set misclassification rates: h = 0: 17.9%; h = 2: 19.6%; h = 5: 20.2% and h = 10: 22.9%. There is no evidence from this study that ANNs outperform conventional methods of analysing epidemiological data. Increasing the complexity of the models serves only to overfit the model to the data. It is important that a validation or test data set is used to assess the performance of highly complex ANNs to avoid overfitting. [source] Misclassification rates, critical values and size of the design in measurement systems capability studiesAPPLIED STOCHASTIC MODELS IN BUSINESS AND INDUSTRY, Issue 5 2009D. Zappa Abstract Measurement systems capability analysis aims to test if the variability of a measurement system is small relative to the variability of a monitored process. At present some open questions are related both to the interpretation of the critical values of the indices typically used by practitioners to assess the capability of a gauge and to the choice of the size of the experimental design to test the repeatability and the reproducibility of the measurement process. In this paper, starting from the misclassification rates of a measurement system, we present a solution to these issues. Copyright © 2009 John Wiley & Sons, Ltd. [source] Using BiowinÔ, Bayes, and batteries to predict ready biodegradabilityENVIRONMENTAL TOXICOLOGY & CHEMISTRY, Issue 4 2004Robert S. Boethling Abstract Wether or not a given chemical substance is readily biodegradable is an important piece of information in risk screening for both new and existing chemicals. Despite the relatively low cost of Organization for Economic Cooperation and Development tests, data are often unavailable and biodegradability must be estimated. In this paper, we focus on the predictive value of selected BiowinÔ models and model batteries using Bayesian analysis. Posterior probabilities, calculated based on performance with the model training sets using Bayes' theorem, were closely matched by actual performance with an expanded set of 374 premanufacture notice (PMN) substances. Further analysis suggested that a simple battery consisting of Biowin3 (survey ultimate biodegradation model) and Biowin5 (Ministry of International Trade and Industry [MITI] linear model) would have enhanced predictive power in comparison to individual models. Application of the battery to PMN substances showed that performance matched expectation. This approach significantly reduced both false positives for ready biodegradability and the overall misclassification rate. Similar results were obtained for a set of 63 pharmaceuticals using a battery consisting of Biowin3 and Biowin6 (MITI nonlinear model). Biodegradation data for PMNs tested in multiple ready tests or both inherent and ready biodegradation tests yielded additional insights that may be useful in risk screening. [source] Comparison of surrogate and direct measurement of insulin resistance in chronic hepatitis C virus infection: Impact of obesity and ethnicity,HEPATOLOGY, Issue 1 2010Khoa D. Lam Studies using surrogate estimates show high prevalence of insulin resistance in hepatitis C infection. This study prospectively evaluated the correlation between surrogate and directly measured estimates of insulin resistance and the impact of obesity and ethnicity on this relationship. Eighty-six nondiabetic, noncirrhotic patients with hepatitis C virus (age = 48 ± 7 years, 74% male, 44% white, 22% African American, 26% Latino, 70% genotype 1) were categorized into normal-weight (body mass index [BMI] < 25, n = 30), overweight (BMI = 25-29.9, n = 38), and obese (BMI , 30, n = 18). Insulin-mediated glucose uptake was measured by steady-state plasma glucose (SSPG) concentration during a 240-minute insulin suppression test. Surrogate estimates included: fasting glucose and insulin, glucose/insulin, homeostasis model assessment (HOMA-IR), quantitative insulin sensitivity check index (QUICKI), insulin (I-AUC) and glucose (G-AUC) area under the curve during oral glucose tolerance test, and the Belfiore and Stumvoll indexes. All surrogate estimates correlated with SSPG, but the magnitude of correlation varied (r = 0.30-0.64). The correlation coefficients were highest in the obese. I-AUC had the highest correlation among all ethnic and weight groups (r = 0.57-0.77). HOMA-IR accounted for only 15% of variability in SSPG in the normal weight group. The common HOMA-IR cutoff of ,3 to define insulin resistance had high misclassification rates especially in the overweight group independent of ethnicity. HOMA-IR > 4 had the lowest misclassification rate (75% sensitivity, 88% specificity). Repeat HOMA-IR measurements had higher within-person variation in the obese (standard deviation = 0.77 higher than normal-weight, 95% confidence interval = 0.25-1.30, P = 0.005). Conclusion: Because of limitations of surrogate estimates, caution should be used in interpreting data evaluating insulin resistance especially in nonobese, nondiabetic patients with HCV. HEPATOLOGY 2010 [source] Hybrid Bayesian networks: making the hybrid Bayesian classifier robust to missing training dataJOURNAL OF CHEMOMETRICS, Issue 5 2003Nathaniel A. Woody Abstract Many standard classification methods are incapable of handling missing values in a sample. Instead, these methods must rely on external filling methods in order to estimate the missing values. The hybrid network proposed in this paper is an extension of the hybrid classifier that is robust to missing values. The hybrid network is produced by performing empirical Bayesian network structure learning to create a Bayesian network that retains its classification ability in the presence of missing data in both training and test cases. The performance of the hybrid network is measured by calculating a misclassification rate when data are removed from a dataset. These misclassification curves are then compared against similar curves produced from the hybrid classifier and from a classification tree. Copyright © 2003 John Wiley & Sons, Ltd. [source] Supervised classification and tunnel visionAPPLIED STOCHASTIC MODELS IN BUSINESS AND INDUSTRY, Issue 2 2005David J. Hand Abstract In recent decades many highly sophisticated methods have been developed for supervised classification. These developments involve complex models requiring complicated iterative parameter estimation schemes, and can achieve unprecedented performance in terms of misclassification rate. However, in focusing efforts on the single performance criterion of misclassification rate, researchers have abstracted the problem beyond the bounds of practical usefulness, to the extent that the supposed performance improvements are irrelevant in comparison with other factors influencing performance. Examples of such factors are given. An illustration is provided of a new method which, for the particular problem of credit scoring, improves a relevant measure of classification performance while maintaining interpretability. Copyright © 2005 John Wiley & Sons, Ltd. [source] Bootstrap methods for assessing the performance of near-infrared pattern classification techniquesJOURNAL OF CHEMOMETRICS, Issue 5 2002Brandye M. Smith Abstract Two parametric bootstrap techniques were applied to near-infrared (NIR) pattern classification models for two classes of microcrystalline cellulose, Avicel® PH101 and PH102, which differ only in particle size. The development of pattern classification models for similar substances is difficult, since their characteristic clusters overlap. Bootstrapping was used to enlarge small test sets for a better approximation of the overlapping area of these nearly identical substances, consequently resulting in better estimates of misclassification rates. A bootstrap that resampled the residuals, referred to as the outside model space bootstrap in this paper, and a novel bootstrap that resampled principal component scores, referred to as the inside model space bootstrap, were studied. A comparison revealed that classification rates for both bootstrap techniques were similar to the original test set classification rates. The bootstrap method developed in this study, which resampled the principal component scores, was more effective for estimating misclassification volumes than the residual-resampling method. Copyright © 2002 John Wiley & Sons, Ltd. [source] Using unlabelled data to update classification rules with applications in food authenticity studiesJOURNAL OF THE ROYAL STATISTICAL SOCIETY: SERIES C (APPLIED STATISTICS), Issue 1 2006Nema Dean Summary., An authentic food is one that is what it purports to be. Food processors and consumers need to be assured that, when they pay for a specific product or ingredient, they are receiving exactly what they pay for. Classification methods are an important tool in food authenticity studies where they are used to assign food samples of unknown type to known types. A classification method is developed where the classification rule is estimated by using both the labelled and the unlabelled data, in contrast with many classical methods which use only the labelled data for estimation. This methodology models the data as arising from a Gaussian mixture model with parsimonious covariance structure, as is done in model-based clustering. A missing data formulation of the mixture model is used and the models are fitted by using the EM and classification EM algorithms. The methods are applied to the analysis of spectra of food-stuffs recorded over the visible and near infra-red wavelength range in food authenticity studies. A comparison of the performance of model-based discriminant analysis and the method of classification proposed is given. The classification method proposed is shown to yield very good misclassification rates. The correct classification rate was observed to be as much as 15% higher than the correct classification rate for model-based discriminant analysis. [source] A hierarchical Bayesian model for predicting the functional consequences of amino-acid polymorphismsJOURNAL OF THE ROYAL STATISTICAL SOCIETY: SERIES C (APPLIED STATISTICS), Issue 1 2005Claudio J. Verzilli Summary., Genetic polymorphisms in deoxyribonucleic acid coding regions may have a phenotypic effect on the carrier, e.g. by influencing susceptibility to disease. Detection of deleterious mutations via association studies is hampered by the large number of candidate sites; therefore methods are needed to narrow down the search to the most promising sites. For this, a possible approach is to use structural and sequence-based information of the encoded protein to predict whether a mutation at a particular site is likely to disrupt the functionality of the protein itself. We propose a hierarchical Bayesian multivariate adaptive regression spline (BMARS) model for supervised learning in this context and assess its predictive performance by using data from mutagenesis experiments on lac repressor and lysozyme proteins. In these experiments, about 12 amino-acid substitutions were performed at each native amino-acid position and the effect on protein functionality was assessed. The training data thus consist of repeated observations at each position, which the hierarchical framework is needed to account for. The model is trained on the lac repressor data and tested on the lysozyme mutations and vice versa. In particular, we show that the hierarchical BMARS model, by allowing for the clustered nature of the data, yields lower out-of-sample misclassification rates compared with both a BMARS and a frequen-tist MARS model, a support vector machine classifier and an optimally pruned classification tree. [source] SNPs in ecological and conservation studies: a test in the Scandinavian wolf populationMOLECULAR ECOLOGY, Issue 2 2005J. M. SEDDON Abstract Single nucleotide polymorphisms (SNPs) have the potential to become the genetic marker of choice in studies of the ecology and conservation of natural populations because of their capacity to access variability across the genome. In this study, we provide one of the first demonstrations of SNP discovery in a wild population in order to address typical issues of importance in ecology and conservation in the recolonized Scandinavian and neighbouring Finnish wolf Canis lupus populations. Using end sequence from BAC (bacterial artificial chromosome) clones specific for dogs, we designed assays for 24 SNP loci, 20 sites of which had previously been shown to be polymorphic in domestic dogs and four sites were newly identified as polymorphic in wolves. Of the 24 assayed loci, 22 SNPs were found to be variable within the Scandinavian population and, importantly, these were able to distinguish individual wolves from one another (unbiased probability of identity of 4.33 × 10,8), providing equivalent results to that derived from 12 variable microsatellites genotyped in the same population. An assignment test shows differentiation between the Scandinavian and neighbouring Finnish wolf populations, although not all known immigrants are accurately identified. An exploration of the misclassification rates in the identification of relationships shows that neither 22 SNP nor 20 microsatellite loci are able to discriminate across single order relationships. Despite the remaining obstacle of SNP discovery in nonmodel organisms, the use of SNPs in ecological and conservation studies is encouraged by the advent of large scale screening methods. Furthermore, the ability to amplify extremely small fragments makes SNPs of particular use for population monitoring, where faecal and other noninvasive samples are routinely used. [source] Artificial neural networks as statistical tools in epidemiological studies: analysis of risk factors for early infant wheezePAEDIATRIC & PERINATAL EPIDEMIOLOGY, Issue 6 2004Andrea Sherriff Summary Artificial neural networks (ANNs) are being used increasingly for the prediction of clinical outcomes and classification of disease phenotypes. A lack of understanding of the statistical principles underlying ANNs has led to widespread misuse of these tools in the biomedical arena. In this paper, the authors compare the performance of ANNs with that of conventional linear logistic regression models in an epidemiological study of infant wheeze. Data on the putative risk factors for infant wheeze have been obtained from a sample of 7318 infants taking part in the Avon Longitudinal Study of Parents and Children (ALSPAC). The data were analysed using logistic regression models and ANNs, and performance based on misclassification rates of a validation data set were compared. Misclassification rates in the training data set decreased as the complexity of the ANN increased: h = 0: 17.9%; h = 2: 16.2%; h = 5: 14.9%, and h = 10: 9.2%. However, the more complex models did not generalise well to new data sets drawn from the same population: validation data set misclassification rates: h = 0: 17.9%; h = 2: 19.6%; h = 5: 20.2% and h = 10: 22.9%. There is no evidence from this study that ANNs outperform conventional methods of analysing epidemiological data. Increasing the complexity of the models serves only to overfit the model to the data. It is important that a validation or test data set is used to assess the performance of highly complex ANNs to avoid overfitting. [source] A robust approach for assessing misclassification rates under the two-component measurement error modelAPPLIED STOCHASTIC MODELS IN BUSINESS AND INDUSTRY, Issue 4 2010Daniela Cocchi Abstract The majority of actions designed to improve processes and quality include the assessment of the capability of a measurement system. The statistical model relating the measured value to the true, but not observable, value of a product characteristic is usually Gaussian and additive. In this paper we propose to extend the said model to a more general formulation by introducing the structure of the two-component error model. An approximated method for evaluating the misclassification rates under the two-component error model is proposed and assessed. Copyright © 2009 John Wiley & Sons, Ltd. [source] Misclassification rates, critical values and size of the design in measurement systems capability studiesAPPLIED STOCHASTIC MODELS IN BUSINESS AND INDUSTRY, Issue 5 2009D. Zappa Abstract Measurement systems capability analysis aims to test if the variability of a measurement system is small relative to the variability of a monitored process. At present some open questions are related both to the interpretation of the critical values of the indices typically used by practitioners to assess the capability of a gauge and to the choice of the size of the experimental design to test the repeatability and the reproducibility of the measurement process. In this paper, starting from the misclassification rates of a measurement system, we present a solution to these issues. Copyright © 2009 John Wiley & Sons, Ltd. [source] Shrinkage-based Diagonal Discriminant Analysis and Its Applications in High-Dimensional DataBIOMETRICS, Issue 4 2009Herbert Pang Summary High-dimensional data such as microarrays have brought us new statistical challenges. For example, using a large number of genes to classify samples based on a small number of microarrays remains a difficult problem. Diagonal discriminant analysis, support vector machines, and,k -nearest neighbor have been suggested as among the best methods for small sample size situations, but none was found to be superior to others. In this article, we propose an improved diagonal discriminant approach through shrinkage and regularization of the variances. The performance of our new approach along with the existing methods is studied through simulations and applications to real data. These studies show that the proposed shrinkage-based and regularization diagonal discriminant methods have lower misclassification rates than existing methods in many cases. [source] | ||||||||||||||||