Re: example41:mining sequential patterns
Date: September 17, 2012 11:19AM
Hi Mavro,
Welcome to the forum.
I will explain to you why.
In the dataset, the timestamps are absolute timestamps. For example, in the first sequence of the dataset, {1} appeared at time 0. It was followed by the itemset {1, 2, 3} at time 1. Finally, itemset {1 3} appeared at time 2.
In the sequential patterns, the timestamps are relative. The timestamp for the first itemset of a sequential pattern is always 0. The timestamps for the following itemsets are relative to the first itemset of the sequential pattern. Let me give you an example. Consider the following pattern:
S16 (0, 1 2), (1, 1) 75 %
This patterns does not mean that {1,2} appeared at time 0. It only means that {1,2} appeared together at any time. Then, 1 time unit after, the item 1 appeared. Those a relative timestamps.
This means that pattern S16 would be included in this sequence:
(0, 1), (1, 1 2 3}), (2, 1 3)
But it would also be included in this sequence:
(20, 1), (21, 1 2 3}), (22, 1 3)
because the importance for sequential patterns is the relative timestamps (not the original timestamps).
Hope this is clear.
Best,
Philippe
Edited 1 time(s). Last edit at 09/17/2012 11:23AM by webmasterphilfv.