Mining Data Streams using Option Trees

In today's applications, evolving data streams are stored as very large databases; the databases which grow without limit at a rate of several million records per day. Data streams are ubiquitous and have become an important research topic in the last two decades. Mining these continuous data streams brings unique opportunities, but also new challenges. For their predictive nonparametric analysis, Hoeffding-based trees are often a method of choice, which offers a possibility of any-time predictions. Although one of their main problems is the delay in learning progress due to the presence of equally discriminative attributes. Options are a natural way to deal with this problem. In this paper, Option trees which build upon regular trees is presented by adding splitting options in the internal nodes to improve accuracy, stability and reduce ambiguity. Results based on accuracy and processing speed of algorithm under various memory limits is presented. The accuracy of Hoeffding Option tree with Hoeffding trees under circumstantial conditions is compared.