The method of defensive forecasting is applied to the problem of prediction with expert advice for binary outcomes. It turns out that defensive forecasting is not only competitive with the Aggregating Algorithm but also handles the case of “second-guessing” experts, whose advice depends on the learner’s prediction; this paper assumes that the dependence on the learner’s prediction is continuous.

RÉSUMÉ. La caractérisation est une tâche supervisée de fouille de données qui permet de résumer de manière succincte et concise un ensemble de données. Cette tâche est intéressante dans la mesure où elle ne nécessite pas de contre exemples. Nous proposons un cadre général pour la caractérisation d’un ensemble d’objets, appelé ensemble ’cible’, en nous basant non seulement sur leurs propriétés p...

The diversity of an ensemble can be calculated in a variety of ways. Here a diversity metric and a means for altering the diversity of an ensemble, called “thinning”, are introduced. We experiment with thinning algorithms evaluated on ensembles created by several techniques on 22 publicly available datasets. When compared to other methods, our percentage correct diversity measure algorithm show...

The existing techniques of forecasting a future count either treat the time series of counts as a Gaussian time series or use a random effects based dynamic Poisson model. The normality based approach may not yield valid forecasting, whereas the random effects based model usually generates a complex correlation structure for the time series of counts which may be impractical to use for forecast...

Accurate electricity price forecasting plays an important role in the profits of electricity market participants and the healthy development of electricity market. However, the electricity price time series hold the characteristics of volatility and randomness, which make it quite hard to forecast electricity price accurately. In this paper, a novel hybrid model for electricity price forecastin...

This paper describes three software applications for distribution network load forecasting in a Smart Grid environment: (i) short-term feeder load forecasting, (ii) shortterm substation transformer load forecasting and transformer rating, and (iii) next-year load pocket forecasting. The shortterm feeder load forecasting allows a utility to reduce the possibility of feeder overloading. The subst...

It has been deemed as an effective tool of forecasting performance improvement to combine different component forecasting models. However, current nonlinear combining models are not able to meet the requirement of high forecasting accuracy in practice. To tackle this challenge, this paper constructs a hybrid, named genetic programming and least squared estimation based nonlinear combining metho...

The CMIP3 multi-model ensemble has been widely utilised for climate research and prediction, but the properties and behavior of the ensemble are not yet fully understood. Here we present some investigations into various aspects of the ensemble’s behaviour, in particular focussing on the performance of the multi-model mean. We present an explanation of this phenomenon in the context of the stati...

We apply the method of determinants to study the distribution of the largest singular values of large real rectangular random matrices with independent Cauchy entries. We show that statistical properties of the largest singular values are different from the Tracy-Widom law. Among other corollaries of our method we show an interesting connection between the mathematical expectations of the deter...

Tree-based ensemble models are heavy memorywise. An undesired state of affairs considering nowadays datasets, memory-constrained environment and fitting/prediction times. In this paper, we propose the Globally Induced Forest (GIF) to remedy this problem. GIF is a fast prepruning approach to build lightweight ensembles by iteratively deepening the current forest. It mixes local and global optimi...