CARTscans: A Tool for Visualizing Complex Models

We present CARTscans, a graphical tool that displays predicted values across a fourdimensional subspace. We show how these plots are useful for understanding the structure and relationships between variables in a wide variety of models, including (but not limited to) regression trees, ensembles of trees, and linear regressions with varying degrees of interactions. In addition, the common visualization framework allows diverse complex models to be visually compared in a way that illuminates the similarities and differences in the underlying methods, facilitates the choice of a particular model structure, and provides a useful check for implausible predictions of future observations in regions with little or no data.     

Improved customer choice predictions using ensemble methods

In this paper various ensemble learning methods from machine learning and statistics are considered and applied to the customer choice modeling problem. The application of ensemble learning usually improves the prediction quality of flexible models like decision trees and thus leads to improved predictions. We give experimental results for two real-life marketing datasets using decision trees, ensemble versions of decision trees and the logistic regression model, which is a standard approach for this problem. The ensemble models are found to improve upon individual decision trees and outperform logistic regression.     

Classifying G-protein coupled receptors with bagging classification tree

G-protein coupled receptors (GPCRs) play a key role in different biological processes, such as regulation of growth, death and metabolism of cells. They are major therapeutic targets of numerous prescribed drugs. However, the ligand specificity of many receptors is unknown and there is little structural information available. Bioinformatics may offer one approach to bridge the gap between sequence data and functional knowledge of a receptor. In this paper, we use a bagging classification tree algorithm to predict the type of the receptor based on its amino acid composition. The prediction is performed for GPCR at the sub-family and sub-sub-family level. In a cross-validation test, we achieved an overall predictive accuracy of 91.1% for GPCR sub-family classification, and 82.4% for sub-sub-family classification. These results demonstrate the applicability of this relative simple method and its potential for improving prediction accuracy.     

Decision tree SAR models for developmental toxicity based on an FDA/TERIS database

Humans are exposed to thousands of environmental chemicals for which no developmental toxicity information is available. Structure-activity relationships (SARs) are models that could be used to efficiently predict the biological activity of potential developmental toxicants. However, at this time, no adequate SAR models of developmental toxicity are available for risk assessment. In the present study, a new developmental database was compiled by combining toxicity information from the Teratogen Information System (TERIS) and the Food and Drug Administration (FDA) guidelines. We implemented a decision tree modeling procedure, using Classification and Regression Tree software and a model ensemble approach termed bagging. We then assessed the empirical distributions of the prediction accuracy measures of the single and ensemble-based models, achieved by repeating our modeling experiment many times by repeated random partitioning of the working database. The decision tree developmental SAR models exhibited modest prediction accuracy. Bagging tended to enhance the accuracy of prediction. Also, the model ensemble approach reduced the variability of prediction measures compared to the single model approach. Further research with data derived from animal species- and endpoint-specific components of an extended and refined FDA/TERIS database has the potential to derive SAR models that would be useful in the developmental risk assessment of the thousands of untested chemicals.     

The Generalization Paradox of Ensembles

Deconvolution is usually regarded as one of the ill-posed problems in applied mathematics if no constraints on the unknowns are assumed. This article discusses the idea of well-defined statistical models being a counterpart of the notion of well-posedness. We show that constraints on the unknowns such as positivity and sparsity can go a long way towards overcoming the ill-posedness in deconvolution. We show how these issues are dealt with in a parametric deconvolution model introduced recently. From the same perspective we take a fresh look at two familiar deconvolvers: the widely used Jansson method and another one that minimizes the Kullback-Leibler divergence between observations and fitted values. In the latter case, we point out that in the context of deconvolution and the general linear inverse problems with positivity contraints, a counterpart of the EM algorithm exists for the problem of minimizing the Kullback-Leibler divergence. We graphically compare the performance of these deconvolvers using data simulated from a spike-convolution model and DNA sequencing data.     

基于决策树的异常高频心电图识别

运用决策树技术对高频心电图进行分类和识别. 采用Bagging和Boosting方法,分类准确率得到一定程度的提高,从76.3%分别升至79.8%和77.5%. 采用代价敏感决策树,异常心电图的识别率得到明显提高, 达到75.0%,但是分类准确率大幅度下降为53.2%,正常心电图的识别率急剧恶化,从93.2%降为42.7%. 通过实例选择优化训练集,重新构建决策树,分类准确率和异常心电图的识别率均得到明显提高,分别升至84.4%和73.2%. 与BP神经网络、RBF神经网络、支持向量机的分类实验结果相比,基于优化训练集建立的决策树性能最好. 从该树提取相应的诊断规则,可以提高诊断的准确度.     

Classification ensembles for unbalanced class sizes in predictive toxicology

This paper investigates the effects of the ratio of positive-to-negative samples on the sensitivity, specificity, and concordance. When the class sizes in the training samples are not equal, the classification rule derived will favor the majority class and result in a low sensitivity on the minority class prediction. We propose an ensemble classification approach to adjust for differential class sizes in a binary classifier system. An ensemble classifier consists of a set of base classifiers; its prediction rule is based on a summary measure of individual classifications by the base classifiers. Two re-sampling methods, augmentation and abatement, are proposed to generate different bootstrap samples of equal class size to build the base classifiers. The augmentation method balances the two class sizes by bootstrapping additional samples from the minority class, whereas the abatement method balances the two class sizes by sampling only a subset of samples from the majority class. The proposed procedure is applied to a data set to predict estrogen receptor binding activity and to a data set to predict animal liver carcinogenicity using SAR (structure-activity relationship) models as base classifiers. The abatement method appears to perform well in balancing sensitivity and specificity.     

When can we improve on sample average approximation for stochastic optimization?

We explore the performance of sample average approximation in comparison with several other methods for stochastic optimization. The methods we evaluate are (a) bagging; (b) kernel density estimation; (c) maximum likelihood estimation; and (d) a Bayesian approach. We use two test sets: first a set of quadratic objective functions allowing different types of interaction between the random component and the univariate decision variable; and second a set of portfolio optimization problems. We make recommendations for effective approaches.