Updated for Spark 1.2 - fails to overwrite qvalues - and other minor changes

[edi-bice]

Updated for Spark 1.2 as bundled with Cloudera 5.3.3 and fixed issue with saveAsTextFile failing to overwrite qvalues - now saving each level's qvalues separately. Also changed to allow parsing input with the third column being non-numeric (ignores). Added option to specify serialization.

[Auroratan]

I found a big problem in your louvain algorithm code, only the karate data set is appropriate, however, the results of other data sets are quite different from the authenticity, and what should I do. @ctwardy

[ctwardy]

@Auroratan - the Louvain code is not mine. I haven't used this project.

[Auroratan]

I'm sorry to hear that you can not solve my problem
Do you have any other ideas in the community to detect algorithms that improve community consolidation for module accuracy and member accuracy @ctwardy

[ctwardy]

@Auroratan Perhaps ask someone on this project? I see contributions by @jgawrilo, @eric-kimbrel, and @edi-bice. Perhaps one of them can provide a definitive answer and close the thread. Poking around, it seems like this may have been deprecated in favor of https://github.com/Sotera/distributed-graph-analytics, which has activity through January 2016.

(Also, can you direct message me about why you wrote to me? If I appeared to be connected to this project then I need to fix our team Github configuration.)