Mastering Machine Learning with Spark 2.x

The sequential pattern mining problem

Let's move on to formalizing, the third and last pattern matching question we tackle in this chapter. Let's look at sequences in more detail. A sequence is different from the transactions we looked at before in that the order now matters. For a given item set I, a sequence S in I of length l is defined as follows:

s = <s_1, s₂, ..., s_l>

Here, each individual s_i is a concatenation of items, that is, s_i = (a_i1 ... a_im), where a_ij is an item in I. Note that we do care about the order of sequence items s_i but not about the internal ordering of the individual a_ij in s_i. A sequence database S consists of pairs of sequence IDs and sequences, analogous to what we had before. An example of such a database can be ...