Exploring the MPdist distance
MPdist offers a way to calculate the distance between two time series. Strictly speaking, the MPdist distance is a distance measure that is based on the Matrix Profile. It is much slower to compute than the Euclidean distance, but it does not require the time series to have the same size.
As you might expect, it must offer many advantages when compared to the Euclidean distance, as well as other existing distance metrics. The main advantages of MPdist, according to the people that created it, are the following:
- It is more flexible regarding the way it compares data than most existing distance functions.
- It considers similarities of data that may not take place at the same time, where time means at the same index.
- MPdist is considered more robust in specific analytics scenarios due to the way it is computed. More specifically, MPdist is more robust to spikes and missing values.
As MPdist is based on the Matrix Profile, calculating...
