Classification and dynamic class detection of real time data for tsunami warning system

Advances in data storage technology have led to the ability to store the data for real-time transactions. Such processes lead to data which often grow without limit and are referred to as data streams. In addition, the development of sensor technology has resulted in the possibility of measuring many events in real time. While data mining has become a fairly well established field now, the data stream problem poses a number of unique challenges which are not easily solvable by traditional mining methods. In a traditional data mining classification task, it is assumed that the total number of classes is fixed. This assumption may not be valid in a real streaming environment, where new classes may evolve at any time. Another problem ignored by the existing data stream techniques is Concept-drift, which occurs in the stream when the underlying concept of the data changes over time. Thus, the classification model must be designed to reflect the most recent concept. The proposed system aims to enhance the existing Tsunami warning system by applying data stream mining Techniques to the real-time sea-level data. The sensors record the various sea level reading and pass it to the buoy located at deep sea. Feature extraction is performed for the training data collected from deep sea buoys. Data are observed every 15 seconds and classified using KNN algorithm. Data that are identified as outliers contribute to the automatic detection of novel class in the presence of concept drift. It also deals with the detection of recurring class, which reappears in the stream after a long time interval. Finally the system is designed to generate Tsunami alerts with maps and evacuation routes.