Classification Event Sequences via Compact Big Sequence

Document Type : Original Article


Faculty of Science, Benha University


The sequence classification is considered as one of the important data mining tasks. It has a broad range of real-world applications such as bioinformatics, medicine, finance, and abnormal detection. In the literature, several algorithms have been proposed for sequence classification from different aspects. Existing algorithms can be partitioned into three types feature-based, distance-based, and model-based algorithms. In particular, the feature-based algorithms are widely applied for the sequence classification in the literature. In this paper, we propose a new event sequence classification method that based on the idea of the compact big sequence (BigSeq). Our classification method called CBigSeq. It is feature-based method where the features are the used Big Sequences in our model. To evaluate CBigSeq, we compare it with the feature-based method, SeqDT (the state-of-the-art sequence classification algorithm). Our performance study shows that CBigSeq can achieve better performance than SeqDT with respect to classification accuracy, total response time, and count of utilized patterns.


Main Subjects