Using iSAX to Approximate MPdist

So far in this book, we have seen the use of iSAX for searching subsequences and joining iSAX indexes based on SAX representations but no other applications of it.

In this chapter, we are going to use iSAX indexes to approximately calculate the Matrix Profile vectors as well as the MPdist distance between two time series – we are still going to use iSAX for searching and joining, but the end results are going to be more sophisticated. The idea that governs this chapter is the perception that terminal nodes in an iSAX index group have similar subsequences from a SAX representation perspective – this is what we are trying to take advantage of for our approximate computations.

In this chapter, we are going to cover the following main topics:

• Understanding the Matrix Profile
• Computing the Matrix Profile using iSAX
• Understanding MPdist
• Calculating MPdist using iSAX
• Implementing the MPdist calculation in Python
...