Matrix Profile

#Terms

WHAT IS… ?

The Matrix Profile is a data structure and associated algorithms that help solve the dual problem of anomaly detection and motif discovery. It is robust, scalable and largely parameter-free. The generalization of the matrix profile not only makes it adaptable to different data sources and dimensions but also desirable for other machine learning problems.

HOW IS IT USED ?

Matrix Profile is a two-stage algorithm. In the first stage, Matrix Profile takes a fixed-size sliding window that is displaced over the data in order to quantify the distance between the pattern in the window and the rest of the series. This results in a vector of distances called the matrix profile, which will have minimum values in the motifs of the series and maximum values in the anomalies.