Logo des Repositoriums

Auflistung nach Autor:in "Džeroski, Sašo"

1 - 5 von 5
  • Konferenzbeitrag
    Discovering the most important factors for communities of soil microarthropods using machine learning
    (Sh@ring – EnviroInfo 2004, 2004) Demšar, Damjan; Džeroski, Sašo; Krogh, Paul Henning; Larsen, Thomas
    In agricultural soil a set of anthropogenic events shapes the ecosystem processes and populations. The risk of impact from anthropogenic sources on the soil environment is almost exclusively assessed for chemicals, although in agriculture other factors like crop and tillage have large impact too. Thus, the farming system as a whole should be evaluated and ranked according to its environmental benefits and impacts. Our starting point is the availability of data sets describing the agricultural events and the soil biological parameters. Using that datasets and machine learning methods for inducing regression and model trees, we produced empirically based models useful for predicting the soil quality in terms of quantities describing the soil microarthropod community from agricultural measures. However, inducing models for predicting soil quality is not our only goal. What we are also interested is to discover additional knowledge on a higher level and identify the most important factors for population densities of springtails and mites and their biodiversity. We do that by preferring smaller and simpler models to bigger and more complex models, while trying to minimize the performance loss of the models at the same time. Using that approach we identify that microarthropod communities are most sensitive to crops and tillage.
  • Konferenzbeitrag
    Integrating knowledge-based and data-driven modeling of population dynamics
    (Sustainability in the Information Society, 2001) Džeroski, Sašo; Todorovski, Ljupco
    The paper is concerned with integrating knowledge-based modeling or modeling from first principles, with data-driven or automated modeling of dynamic systems. We propose an approach to representing knowledge about processes in population dynamics domains and a method to transform such knowledge into an operational form that could be used by systems for discovery of differential equations. In this way, we improve the ability of computer systems to exploit both knowledge and data in the process of automated modeling of dynamic systems.
  • Konferenzbeitrag
    Integrating Knowledge-Driven and Data-Driven Approaches to Modeling
    (Sh@ring – EnviroInfo 2004, 2004) Todorovski, Ljupe; Džeroski, Sašo
    In this paper, we present a modeling framework that integrates the knowledge-based theoretical approach to modeling with the data-driven empirical modeling of dynamic systems. The framework allows for integration of modeling knowledge specific to the domain of interest in the process of model induction from measured data. The knowledge is organized around the central notion of basic processes in the domain, their models, and includes guidelines for combining models of individual processes into a model of the entire observed system. We present a method for automated translation of the knowledge into the operational form of grammars that constrain the space of candidate models considered during the induction process. The developed framework is applied to two tasks of modeling dynamic systems from noisy measurement data in the domains of population and hydro dynamics.
  • Konferenzbeitrag
    Relating Biodiversity of River Communities to Physical and Chemical Water Properties
    (Sustainability in the Information Society, 2001) Džeroski, Sašo; Grbovic, Jasna
    We address the problem of finding relationships between the physical and chemical properties of river water and the biodiversity of the community present in that water . We apply the machine learning approach of induction of regression trees to biological and chemical data collected through regular monitoring of rivers in Slovenia. A predictive model is built, which identifies the most important parameters for predicting the species richness (the number oftaxa) of the community: these include biological oxygen demand (an overall indicator of pollution), water temperature, the season (month), total hardness, NO3, SiO2 and alkalinity.
  • Konferenzbeitrag
    Relating personality traits and mercury exposure in miners with machine learning methods
    (Sh@ring – EnviroInfo 2004, 2004) Ženko, Bernhard; Džeroski, Sašo; Kobal, Alfred B.; Kobal Grum, Darja; Arneri, Niko; Osredkar, Joško; Horvat, Milena
    We use machine learning/data mining methods to analyse scientific data in the area of environmental epidemiology, i.e., the study of the influence of environmental factors on human health. In particular, the aim of this study was an evaluation of the impact of long-term past occupational exposure to elemental mercury vapour (Hg°) on the mental health, i.e., personality traits of ex-mercury miners. Personality traits were defined by the Eysenck Personality Questionnaire (EPQ) and Emotional States Questionnaire (ESQ), which produced scores for traits such as depression and negative self-concept. Statistical analyses were performed to determine if there are significant differences between the values of the scores for ex-miners and controls. For the psychological traits for which significant differences were found between exminers and non-miners, we performed regression analysis. The target variables were the personality trait scores, while the independent/explanatory variables were the indices of previous occupational exposure to Hg°, medical history and lifestyle habits and some biological indices of actual non-occupational exposure. Regression/model trees were used to perform the analyses and revealed many interesting findings, e.g., that alcohol consumption and mercury exposure increase the depression score.