A data-oriented approach for outlier detection

Authors

  • Nripesh Trivedi Department of mathematical sciences. Indian Institute of Technology, Varanasi, India, India

In this paper, characteristics of data obtained from the sensors (used in OpenSense project) are identified in order to build a data-oriented approach. This approach consists of application of Class Outliers: Distance Based (CODB) and Hoeffding tree algorithms. Subsequently, machine learning models were built to detect outliers in a sensor data stream. The approach presented in this paper may be used for developing methodologies for data-oriented outlier detection