In this paper, we present a method based on fuzzy transforms to establish dependencies between numerical attributes in datasets. We find the best fuzzy partitions of the attribute domains with respect to which we determine the direct and inverse fuzzy transforms. We use two specific regression indexes (which must be smaller than a threshold deduced experimentally) for evaluating dependency between numerical attributes. The experiments are conducted on two well known datasets: ''El Nino'' (http://kdd.ics.uci.edu/databases/el_nino/el_nino.data.html) and the remote sensing data determined from US Forest Service (Region 2, Resource Information System, http://kdd.ics.uci.edu/databases/covertype/covertype.data.html). Our results are quite in agreement with the regression analysis of the same data.
Fuzzy transforms method and attribute dependency in data analysis
LOIA, Vincenzo;SESSA, Salvatore
2010-01-01
Abstract
In this paper, we present a method based on fuzzy transforms to establish dependencies between numerical attributes in datasets. We find the best fuzzy partitions of the attribute domains with respect to which we determine the direct and inverse fuzzy transforms. We use two specific regression indexes (which must be smaller than a threshold deduced experimentally) for evaluating dependency between numerical attributes. The experiments are conducted on two well known datasets: ''El Nino'' (http://kdd.ics.uci.edu/databases/el_nino/el_nino.data.html) and the remote sensing data determined from US Forest Service (Region 2, Resource Information System, http://kdd.ics.uci.edu/databases/covertype/covertype.data.html). Our results are quite in agreement with the regression analysis of the same data.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
