| Author | Ronnachai Lorjaroenmongkol |
| Call Number | AIT Thesis no.CS-97-30 |
| Subject(s) | Data mining
|
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of
Science |
| Publisher | Asian Institute of Technology |
| Abstract | Empirical comparisons and evaluations of data mining techniques and algorithms are
often useful to the selection of appropriate tools to perform on specific tasks. One particularly
attractive methodology in KDD and data mining is the theory of rough sets which is a
relatively new research direction concerned with analysis and modeling of classifications and
decision problems involving vague and uncertain information. However, there is a need to
evaluate and characterize the types of practical problems, especially in large databases, to
which the rough sets approach is suitable. The comparative experiments were performed on
both real and public repository data sets. Two experimental series were conducted to evaluate
empirically the behavior and performance of algorithms. The first experiment included a
rough-set-based data mining system (Rosetta) and other two well-known methods named CN2
and PEBLS. The second experiment was to compare among various algorithms for rule
extraction within the framework of rough sets. In this experiment, a number of algorithms for
extracting rules involved Local covering, All global covering, Local certain and possible rules,
and All rules methods. Moreover, two sampling strategies, simple random sampling and
stratified simple random sampling, were employed when producing training data sets. An
appropriate statistical hypothesis testing was used to determine whether there exists a
significant difference between their estimated accuracy. Moreover, the extracted rules were
manually investigated and reported. Finally, strengths and weaknesses of algorithms being
compared were pointed out, and the condition under which a particular algorithm could be
expected to outperform or under-perform the others were indicated. |
| Year | 1997 |
| Type | Thesis |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Phan, Minh Dung |
| Examination Committee(s) | Batanov, Dentcho N.;Devadason, Francis J. |
| Scholarship Donor(s) | Asian Institute of Technology (Partial); |
| Degree | Thesis (M.Sc.) - Asian Institute of Technology, 1997 |