Academia.eduAcademia.edu

Figure 4 – uploaded by Miguel Toro

See full PDFdownloadDownload figure
Fisher data from UCT; 150 recs., 4 attributes, 3 classes.

Table 4 Fisher data from UCT; 150 recs., 4 attributes, 3 classes.

Related Figures (6)

6. Results We have performed experiments with our algorithm on several datasets, and compared results with those published in the literature. Data used to test our classification tool has been gathered from public domain in the web, mainly from the UCI Machine Learning Repository[19]. Unless stated all files have been previously clean. Records with unknown attribute values were removed or converted to a mean value. Using a random function on the complete file each time, the standard ten-fold cross validation procedure was used on all We consider a pattern strong, if the area were it belongs is clearly more disjunctive than others, with respect to a given class. Secondly, the algorithm sees whether patterns p’ and p show the presence of exclusive values. Applying the exclusive function ex() from (16) we can have:
6. Results We have performed experiments with our algorithm on several datasets, and compared results with those published in the literature. Data used to test our classification tool has been gathered from public domain in the web, mainly from the UCI Machine Learning Repository[19]. Unless stated all files have been previously clean. Records with unknown attribute values were removed or converted to a mean value. Using a random function on the complete file each time, the standard ten-fold cross validation procedure was used on all We consider a pattern strong, if the area were it belongs is clearly more disjunctive than others, with respect to a given class. Secondly, the algorithm sees whether patterns p’ and p show the presence of exclusive values. Applying the exclusive function ex() from (16) we can have:
UCL, 303 records, 13 attributes, 5 classes. Values taken from a larger NCU table. 10-fold cross-validation. All missing values replaced by their mean. Values labeled “”? converted to its mean value. Attribute chol used with an interval of two.
UCL, 303 records, 13 attributes, 5 classes. Values taken from a larger NCU table. 10-fold cross-validation. All missing values replaced by their mean. Values labeled “”? converted to its mean value. Attribute chol used with an interval of two.
UCL, 768 records, 8 attributes, 2 classes. Values from [22, 21]. Statlog results taken with 12-fold cross-validation
UCL, 768 records, 8 attributes, 2 classes. Values from [22, 21]. Statlog results taken with 12-fold cross-validation
UCI, 270 records, 13 attributes, 5 classes. From the 303 case Cleveland Hart. All record's missing values replaced by their mean. Source NCU, Poland. Table 1: Statlog Heart disease file.
UCI, 270 records, 13 attributes, 5 classes. From the 303 case Cleveland Hart. All record's missing values replaced by their mean. Source NCU, Poland. Table 1: Statlog Heart disease file.
Training 4435, test 2000 recs. 36 semi-continuous attributs, six classes. We used only four attributes as suggested: 17, 18,19 and 20.
Training 4435, test 2000 recs. 36 semi-continuous attributs, six classes. We used only four attributes as suggested: 17, 18,19 and 20.
UCT, 798 recs., 38 attributes, 5 classes. Results obtained by using Attribute intervals: width= 20, strength = 700, len = 50 and thick = 50 Table 4: Annealing file. (Sterling & Buntine).
UCT, 798 recs., 38 attributes, 5 classes. Results obtained by using Attribute intervals: width= 20, strength = 700, len = 50 and thick = 50 Table 4: Annealing file. (Sterling & Buntine).
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved