next up previous contents
Next: 6.2.1 Number of Rows Up: 6. Characterizing Logistic Regression Previous: 6.1.3 BC   Contents


6.2 Synthetic Datasets

We would like to compare LR-CGEPS, LR-CGDEVEPS, CG-MLE with the classifiers described in Section 6.1 in a controlled environment. To that end we have run experiments on the synthetic datasets of Section 5.1.3.3 to examine how performance is affected by the number of rows, number of attributes, sparsity and attribute coupling of the data. We allowed twenty-four hours per classification task. We do not report any results longer than this period.



Subsections
next up previous contents
Next: 6.2.1 Number of Rows Up: 6. Characterizing Logistic Regression Previous: 6.1.3 BC   Contents
Copyright 2004 Paul Komarek, komarek@cmu.edu