Nearest Neighbor Classification Rule for Mixtures of Discrete and Continuous Random Variables

In this paper very simple nonparametric classification rule for mixtures of discrete and continuous random variables is described. It is based on the method of nearest neighbor proposed by Cover and Hart (1967). The bounds on the limit of the nearest neighbor rule risks are given. Both lower and upper bound depend on the Bayes risk and the loss function. Finally the method is compared with other existing methods on some practical data set.     

Effects of a participatory approach,with systematic impact matrix analysis in herd health planning in organic dairy cattle herds

The animal health and welfare status in European organic dairy production does not in all aspects meet the organic principles and consumers’ expectations and needs to be improved. To achieve this, tailored herd health planning, targeted to the specific situation of individual farms could be of use. The aim of this study was to apply herd health planning in a structured participatory approach, with impact matrix analysis, not previously used in this context, in European organic dairy farms and to assess changes in animal health and welfare. Herd health planning farm visits were conducted on 122 organic dairy farms in France, Germany and Sweden. The farmer, the herd veterinarian and/or an advisor took part in the farm discussions. The researcher served as facilitator. Baseline data on the animal health status of the individual farm, collected from national milk recording schemes, were presented as an input for the discussion. Thereafter a systematic impact matrix analysis was performed. This was to capture the complexity of individual farms with the aim to identify the farm-specific factors that could have a strong impact on animal health. The participants (i.e. farmer, veterinarian and advisor) jointly identified areas in need of improvement, taking the health status and the interconnected farm system components into account, and appropriate actions were jointly identified. The researcher took minutes during the discussions, and these were shared with the participants. No intervention was made by the researcher, and further actions were left with the participants. The number of actions per farm ranged from 0 to 22. The change in mortality, metabolic diseases, reproductive performance and udder health was assessed at two time points, and potential determinators of the change were evaluated with linear regression models. A significant association was seen between change in udder health, as measured by the somatic cell count, and country. At the first follow-up, a significant association was also found between change in the proportion of prolonged calving interval and the farmers’ desire to improve reproductive health as well as with an increase in herd size, but this was not seen at the second follow-up. The degree of implementation of the actions was good (median 67%, lower quartile 40%, upper quartile 83%). To conclude, the degree of implementation was quite high, improvement of animal health could not be linked to the herd health planning approach. However, the approach was highly appreciated by the participants and deserves further study.     

Decision Support through Mass and Energy Flow Management in the Vehicle-Refinishing Sector

In the past few decades, major advances in environmental protection within the coating application industry have been made. In spite of this technological progress, approximately 50% of industrial-solvent emissions still come from the paint-application sector. The advances made in reducing emissions for plants requiring licensing have unfortunately had no influence on the environmental efforts of smaller companies. Solvent-reduced painting systems, such as high-solid paints, water-based coating, and powder coating have not been able to achieve acceptance, nor have innovative application technologies. The principal arguments against a conversion to these ecologically more favorable alternatives were related to cost and quality.
Recently, the EU Solvent Directive (1999/13/EC) went into effect, aiming to significantly reduce industrial-solvent emissions. Up until this point, however, instruments enabling smaller companies to determine their solvent emissions and to simultaneously develop process-improvement potentials while keeping costs in mind have been missing.
Using the mass and energy flow-management approach, cost structures and environmental benefits can be made transparent to the entrepreneur. The primary result of the research projects presented here is the computer-based mass and energy flow model called the individual computer-aided mass and energy flow model for the vehicle-refinishing sector (IMPROVE). It can be used as a detailed business-consultancy tool. Based upon this, practical guidelines were developed for easy orientation and activity planning. They can be used by companies to help them fulfill the requirements of environmental legislation and to display the benefits that can be achieved by various emission-reduction measures.     

The empirical question of thresholds and mechanisms of mate choice

Summary Theoretical discussions concerning how animals might best sample and select mates have suggested that individuals could base decisions either on a sample of mates (sampled-based decisions) or on a threshold of comparison (threshold-based decisions). Recent theoretical work demonstrates that threshold-based mating decisions generate higher expected fitness than sample-based mating decisions when search costs exist. Empirical results from most unmanipulated systems, however, either conclude that females make sample-based decisions or are inconclusive. A few experimental studies designed to detect mating thresholds purport to demonstrate threshold-based choice but an examination of these studies indicates such conclusions were premature. We believe that few examples of threshold-based choice exist because protocols designed to identify mating thresholds were often inconsistent with models of threshold choice. We suggest that future empirical work strive not to document mating thresholdsper se. Rather, future work might best reveal decision rules by manipulating the distribution of quality among potential mates; such manipulations predict uniquely how females using sample-based and threshold-based decision rules should behave.     

Site fidelity in predictable and unpredictable habitats

Summary Site fidelity, the tendency to return to a previously occupied location, has been observed in numerous species belonging to at least three phyla. In this paper I develop a general model using dynamic programming to investigate conditions under which fidelity to a previously occupied territory will be advantageous. The results predict that site fidelity should be inversely related to heterogeneity in territory quality and the animal's lifespan and positively related to the cost of changing territories, age and probability of mortality in the habitat. The predictability of reproductive outcome (defined as the probability that next period's outcome will be the same as this period's outcome) also affects site fidelity. In predictable habitats, changing territories may be favoured after a bad previous outcome. In contrast, settlement should be independent of the previous outcome in unpredictable habitats. Individuals should also be site-faithful in unpredictable habitats, as long as the mean territory quality is equal among available territories. I also investigate the success of two potential decision rules (always stay and win-stay: lose-switch) relative to the optimal settlement strategy. The results show that these rules may perform as well as the optimal strategy under certain conditions. The always stay strategy does well in unpredictable habitats, when the mean quality within a territory is equal among territories. In contrast, the win-stay: lose-switch strategy performs best in predictable habitats.     

Estimating tree-based dynamic treatment regimes using observational data with restricted treatment sequences

A dynamic treatment regime (DTR) is a sequence of decision rules that provide guidance on how to treat individuals based on their static and time-varying status. Existing observational data are often used to generate hypotheses about effective DTRs. A common challenge with observational data, however, is the need for analysts to consider “restrictions” on the treatment sequences. Such restrictions may be necessary for settings where (1) one or more treatment sequences that were offered to individuals when the data were collected are no longer considered viable in practice, (2) specific treatment sequences are no longer available, or (3) the scientific focus of the analysis concerns a specific type of treatment sequences (eg, “stepped-up” treatments). To address this challenge, we propose a restricted tree–based reinforcement learning (RT-RL) method that searches for an interpretable DTR with the maximum expected outcome, given a (set of) user-specified restriction(s), which specifies treatment options (at each stage) that ought not to be considered as part of the estimated tree-based DTR. In simulations, we evaluate the performance of RT-RL versus the standard approach of ignoring the partial data for individuals not following the (set of) restriction(s). The method is illustrated using an observational data set to estimate a two-stage stepped-up DTR for guiding the level of care placement for adolescents with substance use disorder.     

三化螟种群系统的最优管理决策

以三化螟Tryporyza invertulas(Walker)种群动态模型和水稻产量损失预测模型为基础,根据水稻插植期、品种抗性,保护利用自然天敌和杀虫剂多次使用等控制措施以及它们的各种不同组合对该虫种群动态、水稻产量损失串和净收益的影响,以净收益最大为目标函数,研究三化螟种群的最优管理决策。其中,对昆虫种群动态模拟方法作了一点改进,它综合了前人所提出的种群动态模型的优点。建立的系统模型能够提供包括农业防治、生物防治和化学防治措施在内的、对三化螟种群实施有效管理的最优决策方案。