ABSTRACT
In this chapter, we discuss the experiments and evaluation process in detail. We use two di¡erent datasets with di¡erent numbers of instances and class distributions. We compare the features extracted with our approach, namely, the hybrid feature set (HFS), with two other baseline approaches: (1) the binary feature set (BFS), and (2) the derived assembly feature set (DAF). For classi’cation, we compare the performance of three di¡erent classi’ers on each of these feature sets, which are Support Vector Machine (SVM), Naïve Bayes (NB), Bayes Net, decision tree, and boosted decision tree. We show the classi’cation accuracy, false positive and false negative rates for our approach and each of the baseline techniques. We also compare the running times and performance/cost tradeo¡ of our approach compared to the baselines.
