Improved nearest neighbor classifiers by weighting and selection of predictors

Nearest neighborhood classification is a flexible classification method that works under weak assumptions. The basic concept is to use the weighted or un-weighted sums over class indicators of observations in the neighborhood of the target value. Two modifications that improve the performance are considered here. Firstly, instead of using weights that are solely determined by the distances we estimate the weights by use of a logit model. By using a selection procedure like lasso or boosting the relevant nearest neighbors are automatically selected. Based on the concept of estimation and selection, in the second step, we extend the predictor space. We include nearest neighborhood counts, but also the original predictors themselves and nearest neighborhood counts that use distances in sub dimensions of the predictor space. The resulting classifiers combine the strength of nearest neighbor methods with parametric approaches and by use of sub dimensions are able to select the relevant features. Simulations and real data sets demonstrate that the method yields better misclassification rates than currently available nearest neighborhood methods and is a strong and flexible competitor in classification problems.     

Global sourcing in integrated network structures: The case of hybrid purchasing organizations

An issue for companies in globally competitive markets is to adapt their organizational structures and governance in increasingly complex organizations. At the functional level of purchasing, companies are turning to hybrid purchasing organizations in order to leverage global sourcing benefits. One of the key challenges in this context is to distinguish between categories to be integrated across sites and those that are to remain under the authority of each purchasing location in order to maximize purchasing synergies.This paper presents a purchasing portfolio model that provides a comprehensive view of relevant global synergy dimensions. Based on a literature review, a theoretically-grounded purchasing portfolio model for global sourcing is developed. The theoretical basis stems from information processing theory, organizational buying behavior (OBB) and transaction cost economics. The validity of the model is explored by means of case study research. Complementing the extant literature, the theoretical contribution of the paper lies in not only addressing category selection criteria for exploiting economies of scale, but also for economies of information and learning and economies of process. Furthermore, it provides insights regarding integrated network coordination mechanisms at the functional level of purchasing.     

Beliefs About and Attitudes Toward Gambling in French-Speaking Switzerland

The present study aims to analyze attitudes and beliefs of the French-speaking general Swiss population (n = 2500; female n = 1280; mean age = 43 years) as regards gambling, which are to date almost exclusively studied in the North American and Australian contexts. Beliefs related to gambling include the perception of the effectiveness of preventive measures toward gambling, the comparative risk assessment of different addictive behaviors, the perceived risks of different types of gambling and attitudes are related to the gambler’s personality. The general population perceived gambling rather negatively and was conscious of the potential risks of gambling; indeed, 59.0% of the sample identified gambling as an addictive practice. Slot machines were estimated to bear the highest risk. Compared with women and older people, men and young people indicated more positive beliefs about gambling; they perceived gambling as less addictive, supported structural preventive measures less often, and perceived gambling as a less serious problem for society. Gamblers were more likely to put their practices into perspective, perceiving gambling more positively than non-gamblers. General population surveys on such beliefs can deliver insights into preventive actions that should be targeted to young men who showed more favorable views of gambling, which have been shown to be associated with increased risk for problematic gambling.     

Has the HIV Epidemic Peaked?

This study reviews the highly diverse regional and country patterns of HIV epidemics and discusses possible causes of the geographic variation in epidemic sizes. Past trends and projections of the epidemics are presented and the peak years of epidemics are estimated. The potential future impact of new prevention technologies is briefly assessed. A final section summarizes the future impact of the epidemic on key demographic variables. The main finding of this analysis is that the HIV epidemic reached a major turning point over the past decade. The peak years of HIV incidence rates are past for all regions, and the peaks of prevalence rates are mostly in the past except in Eastern Europe, where they are expected to peak in 2008. But owing in part to the life‐prolonging effect of antiretroviral therapy and to sustained population growth, the absolute number of infected individuals is expected to keep growing slowly in sub‐Saharan Africa and to remain near current levels worldwide, thus posing a continuing challenge to public health programs. No country is expected to see a decline in its population size between 2005 and 2050 that is attributable to high mortality related to AIDS.     

Intensive gamblers on German-style slot-machines

Results of German-style slot-machine gambling are reported, stemming from seven individual surveys in representatively sampled households (N=7,643). The aim of the inquiry was the reexamination of studies of the number of gamblers in the population together with gambling hours spent per week by active gamblers. The results are largely identical with already known data. Of persons of voting age or older, 10.2% are active gamblers (4.6 million; having gambled at least once within the last 3 months). Of the active gamblers 0.7% are intensive gamblers (5 hours and more per week). These persons number approximately 32,000 in the Federal Republic of Germany, or from 12,000 to 71,000 (95% confidence interval). In addition to the frequency distributions, interactions with socio-demographic characteristics were examined, and an estimate was made for the prevalence of intensive gamblers with severe stress.This article was first published in Germany: Bühringer, G. & Konstanty, R., (1989) Vielsspieler an Geldautomaten in der Bundesrepublik Deutschland, Suchtgefahren 35, 1–13.     

Comparing a new technology with its predecessor—Steel making

This paper examines the physical and economic performance of a major innovation in steel making, the Basic Oxygen Furnace, and compares it with the performance of the traditional, open-hearth process. Evaluations are based on actual records from five American plants, covering three different time periods, and are set against engineering and management expectations as to the innovation's effects. Interviews with managers and technical personnel complemented the statistical findings. The analysis considers: how the attractiveness of the innovation was determined ex ante; the relative accuracy of the expected technological and economic expectations; the interactions of the innovation with the surrounding production system; and the objectives of post-installation technical and managerial adaptations.     

A comparison of methods for the fitting of generalized additive models

There are several procedures for fitting generalized additive models, i.e. regression models for an exponential family response where the influence of each single covariates is assumed to have unknown, potentially non-linear shape. Simulated data are used to compare a smoothing parameter optimization approach for selection of smoothness and of covariates, a stepwise approach, a mixed model approach, and a procedure based on boosting techniques. In particular it is investigated how the performance of procedures is linked to amount of information, type of response, total number of covariates, number of influential covariates, and extent of non-linearity. Measures for comparison are prediction performance, identification of influential covariates, and smoothness of fitted functions. One result is that the mixed model approach returns sparse fits with frequently over-smoothed functions, while the functions are less smooth for the boosting approach and variable selection is less strict. The other approaches are in between with respect to these measures. The boosting procedure is seen to perform very well when little information is available and/or when a large number of covariates is to be investigated. It is somewhat surprising that in scenarios with low information the fitting of a linear model, even with stepwise variable selection, has not much advantage over the fitting of an additive model when the true underlying structure is linear. In cases with more information the prediction performance of all procedures is very similar. So, in difficult data situations the boosting approach can be recommended, in others the procedures can be chosen conditional on the aim of the analysis.     

Penalized regression with correlation-based penalty

A new regularization method for regression models is proposed. The criterion to be minimized contains a penalty term which explicitly links strength of penalization to the correlation between predictors. Like the elastic net, the method encourages a grouping effect where strongly correlated predictors tend to be in or out of the model together. A boosted version of the penalized estimator, which is based on a new boosting method, allows to select variables. Real world data and simulations show that the method compares well to competing regularization techniques. In settings where the number of predictors is smaller than the number of observations it frequently performs better than competitors, in high dimensional settings prediction measures favor the elastic net while accuracy of estimation and stability of variable selection favors the newly proposed method.     

Estimation in semiparametric models using an auxiliary model

Two classes of semiparametric and nonparametric mixture models are defined to represent general kinds of prior information. For these models the nonparametric maximum likelihood estimator (NPMLE) of an unknown probability distribution is derived and is shown to be consistent and relative efficient. Linear functionals are used for the estimation of parameters. Their consistency is proved, the gain of efficiency is derived and asymptotical distributions are given.