Did you look at the cogroup transformation or the cartesian transformation ?
Regards, Olivier. Le sam. 2 mai 2015 à 22:01, Franz Chien <franzj...@gmail.com> a écrit : > Hi all, > > Can I group elements in RDD into different groups and let each group share > elements? For example, I have 10,000 elements in RDD from e1 to e10000, and > I want to group and aggregate them by another mapping with size of 2000, > ex: ( (e1,e42), (e1,e554), (e3, e554)…… (2000th group)) > > My first approach was to filter the RDD with mapping rules for 2000 times, > and then union them together. However, it ran forever. Does SPARK provide a > way to group elements in RDD like this please? > > > Thanks, > > > Franz >
