Simple technique for detection of outliers in one-dimensional numerical data used for point out anomalous consumption

Abstract

This paper presents a simple technique for detection of outliers in one-dimensional numerical data. This was developed to point out anomaly in the consumption information in the DSO’s database. This, in turn, was inspired from concepts like DBSCAN and K-Means grouping techniques. It has been shown plausible for data whose distribution curve is skewed, positively or negatively.