The paper considers particular interestingness measures, called confirmation measures (also known as Bayesian confirmation measures), used for the evaluation of “if evidence, then hypothesis” rules. The agreement of such measures with a statistically sound (significant) dependency between the evidence and the hypothesis in data is thoroughly investigated. The popular confirmation measures were not defined to possess such form of agreement. However, in error-prone environments, potential lack of agreement may lead to undesired effects, e.g. when a measure indicates either strong confirmation or strong disconfirmation, while in fact there is only weak dependency between the evidence and the hypothesis. In order to detect and prevent such situations, the paper employs a coefficient allowing to assess the level of dependency between the evidence and the hypothesis in data, and introduces a method of quantifying the level of agreement (referred to as a concordance) between this coefficient and the measure being analysed. The concordance is characterized and visualised using specialized histograms, scatter-plots, etc. Moreover, risk-related interpretations of the concordance are introduced. Using a set of 12 confirmation measures, the paper presents experiments designed to establish the actual concordance as well as other useful characteristics of the measures.
The paper presents visualization techniques for interestingness measures. The process of measure visualization provides useful insights into different domain areas of the visualized measures and thus effectively assists their comprehension and selection for different knowledge discovery tasks. Assuming a common domain form of the visualized measures, a set of contingency tables, which consists of all possible tables having the same total number of observations, is constructed. These originally four-dimensional data may be effectively represented in three dimensions using a tetrahedron-based barycentric coordinate system. At the same time, an additional, scalar function of the data (referred to as the operational function, e.g., any interestingness measure) may be rendered using colour. Throughout the paper a particular group of interestingness measures, known as confirmation measures, is used to demonstrate the capabilities of the visualization techniques. They cover a wide spectrum of possibilities, ranging from the determination of specific values (extremes, zeros, etc.) of a single measure, to the localization of pre-defined regions of interest, e.g., such domain areas for which two or more measures do not differ at all or differ the most.
The paper considers a particular group of rule interestingness measures, called Bayesian confirmation measures, which have become the subject of numerous, but often exclusively theoretical studies. To assist and enhance their analysis in real-life situations, where time constraints may impede conducting such time consuming procedures, a visual technique has been introduced and described in this paper. It starts with an exhaustive and non-redundant set of contingency tables, which consists of all possible tables having the same number of observations. These data, originally 4-dimensional, may, owing to an inherent constraint, be effectively represented as a 3-dimensional tetrahedron, while an additional, scalar function of the data (e.g. a confirmation measure) may be rendered using colour. Dedicated analyses of particular colour patterns on this tetrahedron allow to promptly perceive particular properties of the visualized measures. To illustrate the introduced technique, a set of 12 popular confirmation measures has been selected and visualized. Additionally, a set of 9 popular properties has been chosen and the visual interpretations of the measures in terms of the properties have been presented.
4
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
The paper focuses on Bayesian confirmation measures used for evaluation of rules induced from data. To distinguish between many confirmation measures, their properties are analyzed. The article considers a group of symmetry properties. We demonstrate that the symmetry properties proposed in the literature focus on extreme cases corresponding to entailment or refutation of the rule's conclusion by its premise, forgetting intermediate cases. We conduct a thorough analysis of the symmetries regarding that the confirmation should express how much more probable the rule's hypothesis is when the premise is present rather than when the negation of the premise is present. As a result we point out which symmetries are desired for Bayesian confirmation measures. Next, we analyze a set of popular confirmation measures with respect to the symmetry properties and other valuable properties, being monotonicity M, Ex1 and weak Ex1, logicality L and weak L. Our work points out two measures to be the most meaningful ones regarding the considered properties.
5
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
One of the crucial problems in the field of knowledge discovery is development of good interestingness measures for evaluation of the discovered patterns. In this paper, we consider quantitative, objective interestingness measures for "if..., then... " association rules. We focus on three popular interestingness measures, namely rule interest function of Piatetsky-Shapiro, gain measure of Fukuda et al., and dependency factor used by Pawlak. We verify whether they satisfy the valuable property M of monotonic dependency on the number of objects satisfying or not the premise or the conclusion of a rule, and property of hypothesis symmetry (HS). Moreover, analytically and through experiments we show an interesting relationship between those measures and two other commonly used measures of rule support and anti-support.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.