Abstract
Most of the existing data mining approaches to time series prediction use as training data an embed of the most recent values of the time series, following the traditional linear auto-regressive methodologies. However, in many time series prediction tasks the alternative approach that uses derivative features constructed from the raw data with the help of domain theories can produce significant prediction accuracy improvements. This is particularly noticeable when the available data includes multivariate information although the aim is still the prediction of one particular time series. This latter situation occurs frequently in financial time series prediction. This paper presents a method of feature construction based on domain knowledge that uses multivariate time series information. We show that this method improves the accuracy of next-day stock quotes prediction when compared with the traditional embed of historical values extracted from the original data. © Springer-Verlag Berlin Heidelberg 2001.

Abstract
The wastewater plumes show to be very difficult to observed in detail. The several studies already conducted exhibit very complex and patchy structures both in vertical and horizontal sections. It is not clear if this plume patchiness is due to physical processes or measurement limitations. Rapid tow-yo sampling is expected to reduce the time variability during and between transects. The AUVs may be a useful instrument to map and detect wastewater plumes. This paper presents several prediction studies using time series files of actual in-situ measurements integrated in a near field model. The model predictions of the plume characteristics at the end of near field support the definition of the best sampling strategy for an AUV monitoring mission in a Portuguese west coast outfall.

Abstract
This paper presents a review of the state of the art on Intelligent Transportation Systems. ITS involves a large number of research areas and, therefore, this paper focus on those we believe to be the most relevant. The main purpose is to study the achievements attained in the last years and to give an overview of possible directions towards future research.

Abstract
Regression trees are models developed to deal with multiple regression data analysis problems. These models fit constants to a set of axes-parallel partitions of the input space defined by the predictor variables. These partitions are described by a hierarchy of logical tests on the input variables of the problem. Several authors have remarked that the preference criteria used to select these tests have a clear preference for what is known as end-cut splits. These splits lead to branches with a few training cases, which is usually considered as counter-intuitive by the domain experts. In this paper we describe an empirical study of the effect of this end-cut preference on a large set of regression domains. The results of this study, carried out for the particular case of least squares regression trees, contradict the prior belief that these type of tests should be avoided. As a consequence of these results, we present a new method to handle these tests that we have empirically shown to have better predictive accuracy than the alternatives that are usually considered in tree-based models. © Springer-Verlag Berlin Heidelberg 2001.

Abstract
For uniformly asymptotically affine (uaa) Markov maps on train tracks, we prove the following type of rigidity result: if a topological conjugacy between them is (uaa) at a point in the train track then the conjugacy is (uaa) everywhere. In particular, our methods apply to the case in which the domains of the Markov maps are Canter sets. We also present similar statements for (uaa:) and C-r Markov families. These results generalize the similar ones of Sullivan and de Faria for C-r expanding circle maps with r > 1 and have useful applications to hyperbolic dynamics on surfaces and laminations.

Abstract
Usability problems associated with virtual environments are a serious obstacle to their successful development. One source of these problems is that virtual environment toolkits provide only a small number of predefined interaction techniques that are expected to be used regardless of context, hence developers are not encouraged to consider interaction. In addition, there are no generally accepted development methodologies for virtual environments. Therefore, even when developers do consider interaction, it is likely to be in an ad hoc fashion driven by technology rather than requirements. If virtual environments are to be useful in a wider context, it is important to provide developers with methods (and tools to support the methods) by which interaction techniques can be systematically designed, tested and refined. In this paper we present the Marigold toolset which supports such a development process. The process begins with a visual specification of the technique being designed. This is requirements centred because it abstracts from implementation issues. Using the toolset, this specification is refined to a prototype implementation so that the technique can be explored in the context of the other elements of the environment. In this way, the developer can verify the technique against requirements in both the specification and prototype. Additionally, because the specification is readily understandable, users can be involved at both stages of the process.