SPADE algorithm for sequential pattern mining, does RapidMiner has ?

phivu

Hi RapidMiner,

 

SPADE (Sequential Pattern Discovery using Equivalence classes) is another algorithm for sequential pattern mining besides GSP, and CM-SPADE is an improved version of SPADE by using co-occurence information. May i know whether RapidMiner has either of SPADE or CM-SPADE ? If not, will you plan to implement these 2, cos i've seen some articles said SPADE is faster than GSP for large datasets?

 

Thank you very much for this information.

 

P/S: below is the links to the papers of SPADE and CM-SPADE:

 

https://link.springer.com/content/pdf/10.1023/A:1007652502315.pdf

 

https://www.philippe-fournier-viger.com/spmf/PAKDD2014_sequential_pattern_mining_CM-SPADE_CM-SPAM.pdf

 

Best Regards,

Phi Vu

yyhuang

Hi @phivu,

 

Good question. SPADE is a faster algorithm, a vertical data format-based sequential pattern method.

From the source code of our Apriori-based GSP operator, which can be found here 

https://github.com/rapidminer/rapidminer-5/tree/master/src/com/rapidminer/operator/learner/associations/gsp

 it is not an implementation of SPADE, but rather a traditional GSP https://en.wikipedia.org/wiki/GSP_algorithm

 

If you can find an R or Python library for SPADE, please share here. You can combine the power by integrating the R/Python scripts into the workflow.

 

YY

phivu

@yyhuang: Thank you for the info!