The Data Mining Forum
This forum is about data mining
, data science
and big data
: algorithms, source code, datasets, implementations, optimizations, etc. You are welcome to post call for papers, data mining job ads, link to source code of data mining algorithms or anything else related to data mining. The forum is hosted by P. Fournier-Viger
. No registration is required to use this forum!
The differences between PPM and AKOM
Date: May 18, 2019 01:32AM
I read your paper about the CPT\CPT+ and I'm sure what are the differences between All-kth-order Markov (AKOM) and Prediction by Partial Matching (PPM).
I thought that PPM keeps a dictionary for each order which counts the occurrences of an element given the previous sub-sequence and then, in the prediction phase, the algorithm tries to match one shorter length sub-sequence until a match is found (i.e., partial matching).
But then, I read this paper:
Brun, A., & Boyer, A. (2009, September). Towards privacy compliant and anytime recommender systems. In International Conference on Electronic Commerce and Web Technologies (pp. 276-287). Springer, Berlin, Heidelberg.
And it looks like the same explanation for the PPM...
Re: The differences between PPM and AKOM
Date: May 20, 2019 06:51PM
From my understanding, these models are very similar.
PPM order 1 only looks at the last event to predict the next event.
PPM order 2 only looks at the last two events to predict the next event.
PPM order 3 only looks at the last three events to predict the next event.
On the other hand, AKOM of order 3 is like combining PPM of order 1,2 and 3 to predict the next event. AKOM will try to predict using the the last three events, and if it fails, it will try using the last two events, and if it fails, it will try using only the last event.
That is what I know about these models.