Detecting Change in Data Streams

This chapter presents a novel method for the detection and estimation of change in a data stream. To provide statistical guarantees on the reliability of detected changes, this method provides meaningful descriptions and quantification of these changes. Detecting changes in a data stream is an important area of research with many applications. The approach assumes that the points in the stream are independently generated, but otherwise makes no assumptions on the nature of the generating distribution. Thus, these techniques work for both continuous and discrete data. A meta-algorithm for change detection in streaming data is also described. The meta-algorithm reduces the problem from the streaming data scenario to the problem of comparing two sample sets. In order to compare the various statistics for nonparametric change detection, it is necessary to use simulated data, so that the changes in generating distributions are known.