Rule Extraction using Machine Learning Classifiers for Complex Event Processing

The large volume of data processing is always challenging for real-time applications. These applications need an optimal framework for handling large scale data and correlating these streams in real time to make better decision making. Complex event processing has emerged as a novel methodology for handling event streams based on atomic events or complex events to find useful patterns by predefined rules.Rules play a major role in these systems and the streams are matched with rules created through a decision tree and machine learning classifier algorithms. In this research work, we propose a complex event processing based framework for rule extraction as well as a comparative analysis of rule-based classifier algorithms for automatic extraction of rules, and since event rules are based on human expertise, sometimes they fail due to a static approach, therefore there is a need for an automatic rule extraction framework.The comparative analysis is performed using a case study of an air quality dataset that outperformed traditional approaches to rule extraction for stream data. Decision tree fetched most number of rules with an accuracy of 99%.The classifier’s learning rate show how efficiently the rule are fetched.