Hi, I am reading source code of FPGrowth in Mahout too. I guess this implementation is base on this paper: paper <http://infolab.stanford.edu/%7Eechang/recsys08-69.pdf> I still have another question that how does this algorithm keep consistency in MapReduce chain? Hopefully I can have more talk with you. :-)
2011/8/3 Vince Wei (jianwei) <[email protected]> > I am looking into the code in the FPGrowth class, I have some questions: > > > > 1. What does the variable conditionalOfCurrentAttribute > > In the growthBottomUp method and the growthTopDown method means? > > > > 2. the header table is traversed in the fpGrowth method, > > why is it traversed again in the growth method? > > > > By the way the FPGrowth implementation is very different from the paper > > Efficient Incremental Mining of Top-K Frequent Closed Itemsets > > Which can be downloaded from > http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.106.3693&rep=re > p1&type=pdf > > Anyone can give me some documents to help me understand the > implementation? Thanks! > > > > > > Sincerely > > > >
