Robust preprocessing for improving angle based outlier detection technique

Outlier detection is an interesting data mining technique, which focuses on finding rare interesting objects in a data set. An outlier is an object which is noticeably distant from the rest of the data. In this paper, a robust preprocessing session consists of robust data normalization and dimensionality reduction employed for enhancing angle based outlier detection technique. Evaluations performed on the synthetic data set indicated the performance and effectiveness of proposed approach.