Reliability measures for local nodes assessment in classification trees

Most of the modem developments with classification trees are aimed at improving their predictive capacity. This article considers a curiously neglected aspect of classification trees, namely the reliability of predictions that come from a given classification tree. In the sense that a node of a tree represents a point in the predictor space in the limit, the aim of this article is the development of localized assessment of the reliability of prediction rules. A classification tree may be used either to provide a probability forecast, where for each node the membership probabilities for each class constitutes the prediction, or a true classification where each new observation is predictively assigned to a unique class. Correspondingly, two types of reliability measure will be derived-namely, prediction reliability and classification reliability. We use bootstrapping methods as the main tool to construct these measures. We also provide a suite of graphical displays by which they may be easily appreciated. In addition to providing some estimate of the reliability of specific forecasts of each type, these measures can also be used to guide future data collection to improve the effectiveness of the tree model. The motivating example we give has a binary response, namely the presence or absence of a species of Eucalypt, Eucalyptus cloeziana, at a given sampling location in response to a suite of environmental covariates, (although the methods are not restricted to binary response data).

Low-density koala (Phascolarctos cinereus) populations in the mulgalands of south-west Queensland. IV. Abundance and conservation status

Urban encroachment on dense, coastal koala populations has ensured that their management has received increasing government and public attention. The recently developed National Koala Conservation Strategy calls for maintenance of viable populations in the wild. Yet the success of this, and other, conservation initiatives is hampered by lack of reliable and generally accepted national and regional population estimates. In this paper we address this problem in a potentially large, but poorly studied, regional population in the State that is likely to have the largest wild populations. We draw on findings from previous reports in this series and apply the faecal standing-crop method (FSCM) to derive a regional estimate of more than 59 000 individuals. Validation trials in riverine communities showed that estimates of animal density obtained from the FSCM and direct observation were in close agreement. Bootstrapping and Monte Carlo simulations were used to obtain variance estimates for our population estimates in different vegetation associations across the region. The most favoured habitat was riverine vegetation, which covered only 0.9% of the region but supported 45% of the koalas. We also estimated that between 1969 and 1995 similar to 30% of the native vegetation associations that are considered as potential koala habitat were cleared, leading to a decline of perhaps 10% in koala numbers. Management of this large regional population has significant implications for the national conservation of the species: the continued viability of this population is critically dependent on the retention and management of riverine and residual vegetation communities, and future vegetation-management guidelines should be cognisant of the potential impacts of clearing even small areas of critical habitat. We also highlight eight management implications.

Determining the effective dimensionality of the genetic variance-covariance matrix

Determining the dimensionality of G provides an important perspective on the genetic basis of a multivariate suite of traits. Since the introduction of Fisher's geometric model, the number of genetically independent traits underlying a set of functionally related phenotypic traits has been recognized as an important factor influencing the response to selection. Here, we show how the effective dimensionality of G can be established, using a method for the determination of the dimensionality of the effect space from a multivariate general linear model introduced by AMEMIYA (1985). We compare this approach with two other available methods, factor-analytic modeling and bootstrapping, using a half-sib experiment that estimated G for eight cuticular hydrocarbons of Drosophila serrata. In our example, eight pheromone traits were shown to be adequately represented by only two underlying genetic dimensions by Amemiya's approach and factor-analytic modeling of the covariance structure at the sire level. In, contrast, bootstrapping identified four dimensions with significant genetic variance. A simulation study indicated that while the performance of Amemiya's method was more sensitive to power constraints, it performed as well or better than factor-analytic modeling in correctly identifying the original genetic dimensions at moderate to high levels of heritability. The bootstrap approach consistently overestimated the number of dimensions in all cases and performed less well than Amemiya's method at subspace recovery.

What determines the finance-growth nexus? Empirical evidence for threshold models

This paper elaborates the notion of balanced'' financial development that is contingent on a country's general level of development. We develop an empirical framework to address this point, referring to threshold regressions and a bootstrap test for structural shift in a growth equation. We find that countries gain less from financial activity, if the latter fails to keep up with or exceeds what would follow from a balanced expansion path. These analyses contribute to the finance and growth literature in providing empirical support for the balanced'' financial development hypothesis.

Use of routine data for determination of the population pharmacokinetics and enteral biovailability of phenytoin in neonates and infants with seizures

Objective: To investigate the population pharmacokinetics and the enteral bioavailability of phenytoin in neonates and infants with seizures. Methods: Data (5 mg kg-1 day-1) from 83 patients were obtained retrospectively from the medical records following written ethical approval. A one-compartment model was fitted to the data using NONMEM with FOCE-interaction. Between-subject variability (BSV) and interoccasion variability (IOV) were modelled exponentially together with a log transform-both-sides exponential residual unexplained variance (RUV) model. Covariates in nested models were screened for significance (X2, 1, 0.01). Model validity was determined by bootstrapping with replacement (N=500 samples) from the dataset. Results: The parameters of final pharmacokinetic were: Clearance (L h-1) = 0.826.(current Weight [kg]/70)0.75.(1+0.0692.(Postnatal age [days]-11)); Volume of distribution (L) = 74.2.(current Weight [kg]/70); Enteral bioavailability = 0.76; Absorption rate constant (h-1) = 0.167. BSV for clearance and volume of distribution were 74.2% and 65.6%, respectively. The IOV in clearance was 54.4%. The RUV was 51.1%. Final model parameters deviated from mean bootstrap estimates by