Outlier Detection in High Dimensional, Spatial and Sequential Data Sets

Of all the data mining techniques, outlier detection seems closest to the definition of “discovering nuggets of information” in large databases. When an outlier is detected, and determined to be genuine, it can provide insights, which can radically change our understanding of the underlying process. The purpose of the research underlying this thesis was to investigate and devise methods to mine for outliers in different type of data sets. In this thesis, we propose: • An efficient sampling based outlier detection method for large high-dimensional data. • A measure, Spatial Local Outlier Measure (SLOM), which captures the local behavior of datum in their spatial neighborhood. With the help of SLOM, we are able to discern local spatial outliers, which are usually missed by global techniques like “three standard deviations away from the mean”. • A unique approach to mine for sequential outliers using Probabilistic Suffix Trees (PST).