It sounds like you are adding the same key to every element, and joining, in order to accomplish a full cartesian join? I can imagine doing it that way would blow up somewhere. There is a cartesian() method to do this maybe more efficiently.
However if your data set is large, this sort of algorithm for computing Kendall's tau is going to be very slow since it's N^2 and would create an unspeakably large shuffle. There are faster algorithms for this statistic. Also consider sampling your data and computing the join over a small sample to estimate the statistic. On Thu, Aug 28, 2014 at 11:15 AM, Yanbo Liang <yanboha...@gmail.com> wrote: > Maybe you can refer sliding method of RDD, but it's right now mllib > private method. > Look at org.apache.spark.mllib.rdd.RDDFunctions. > > > 2014-08-26 12:59 GMT+08:00 Vida Ha <v...@databricks.com>: > >> Can you paste the code? It's unclear to me how/when the out of memory is >> occurring without seeing the code. >> >> >> >> >> On Sun, Aug 24, 2014 at 11:37 PM, Gefei Li <gefeili.2...@gmail.com> >> wrote: >> >>> Hello everyone, >>> I am transplanting a clustering algorithm to spark platform, and I >>> meet a problem confusing me for a long time, can someone help me? >>> >>> I have a PairRDD<Integer, Integer> named patternRDD, which the key >>> represents a number and the value stores an information of the key. And I >>> want to use two of the VALUEs to calculate a kendall number, and if the >>> number is greater than 0.6, then output the two KEYs. >>> >>> I have tried to transform the PairRDD to a RDD<Tuple2<Integer, >>> Integer>>, and add a common key zero to them, and join two together then >>> get a PairRDD<0, Iterable<Tuple2<Tuple2<key1, value1>, Tuple2<key2, >>> value2>>>>, and tried to use values() method and map the keys out, but it >>> gives me an "out of memory" error. I think the "out of memory" error is >>> caused by the few entries of my RDD, but I have no idea how to solve it. >>> >>> Can you help me? >>> >>> Regards, >>> Gefei Li >>> >> >> >
