Date on Master's Thesis/Doctoral Dissertation

12-2006

Document Type

Master's Thesis

Degree Name

M.S.

Department

Computer Engineering and Computer Science

Committee Chair

Elmaghraby, Adel S.

Committee Co-Chair (if applicable)

Eman, Ahmed

Committee Member

Chang, Dar-Jen

Committee Member

Wong, Julius P.

Committee Member

Wahba, Khaled

Author's Keywords

Data mining; Knowledge discovery

Subject

Data mining; Algorithms

Abstract

The area of Knowledge discovery and data mining is growing rapidly. Feature Discretization is a crucial issue in Knowledge Discovery in Databases (KDD), or Data Mining because most data sets used in real world applications have features with continuously values. Discretization is performed as a preprocessing step of the data mining to make data mining techniques useful for these data sets. This thesis addresses discretization issue by proposing a multivariate discretization (MVD) algorithm. It begins withal number of common discretization algorithms like Equal width discretization, Equal frequency discretization, Naïve; Entropy based discretization, Chi square discretization, and orthogonal hyper planes. After that comparing the results achieved by the multivariate discretization (MVD) algorithm with the accuracy results of other algorithms. This thesis is divided into six chapters, covering a few common discretization algorithms and tests these algorithms on a real world datasets which varying in size and complexity, and shows how data visualization techniques will be effective in determining the degree of complexity of the given data set. We have examined the multivariate discretization (MVD) algorithm with the same data sets. After that we have classified discrete data using artificial neural network single layer perceptron and multilayer perceptron with back propagation algorithm. We have trained the Classifier using the training data set, and tested its accuracy using the testing data set. Our experiments lead to better accuracy results with some data sets and low accuracy results with other data sets, and this is subject ot the degree of data complexity then we have compared the accuracy results of multivariate discretization (MVD) algorithm with the results achieved by other discretization algorithms. We have found that multivariate discretization (MVD) algorithm produces good accuracy results in comparing with the other discretization algorithm.

Download

Share

COinS