Feature selection is used to improve performance of learning algorithms by finding a minimal subset of relevant features. Since the process of feature selection is computationally intensive, a trade-off between the quality of the selected subset and the computation time is required. In this paper, we are presenting a novel, anytime algorithm for feature selection, which gradually improves the quality of results by increasing the computation time. The algorithm is interruptible, i.e., it can be stopped at any time and provide a partial subset of selected features. The quality of results is monitored by a new measure: fuzzy information gain. The algorithm performance is evaluated on several benchmark datasets.

Visualization is known to be one of the most efficient data mining approaches. The human eye can capture complex patterns and relationships, along with detecting the outlying (exceptional) cases in a data set. The main limitation of the visual data analysis is its poor scalability: it is hardly applicable to data sets of high dimensionality. We use the concepts of Fuzzy Set Theory to automate the process of human perception. The automated tasks include comparison of frequency distributions, evaluating reliability of dependent variables, and detecting outliers in noisy data. Multiple perceptions (related to different users) can be represented by adjusting the parameters of the fuzzy membership functions. The applicability of automated perceptions is demonstrated on several real-world data sets. Keywords: Data mining, Fuzzy set theory, data visualization, data perception, rule extraction. 1. Introduction Fayyad et al. [3] have defined the process of knowledge discovery in databases (K...

Introduction Causes of death are a constant subject of medical and demographic research. The issues of interest include the change of the leading causes over time, geographical patterns of deadly diseases, life expectancy, tracking diseases to genetic factors, and many others (Blij and Murphy, 1998). The importance of these studies goes beyond the area of medical 1 Department of Industrial Engineering, Tel-Aviv University, Tel-Aviv 69978, Israel research and health care industry. For example, the life insurance companies have to analyze mortality data in determining the terms of their policies. Unfortunately, the tools used in the research on death causality have been so far limited to the standard statistical techniques, like summarization, regression, analysis of variance, etc. In the last decade, more advanced methods of finding patterns in data have become known under the general name of "data mining", and a formal framework for the entire proces