The open-source RHadoop project to integrate R and Hadoop continues apace, with a new version of the rmr package released this week. Changes in this version improve performance when storing and retrieving R objects from Hadoop with a native serialization process, support for equijoins (a MapReduce-style merge) and some new higher-level R functions to make writing map-reduce tasks simpler. The full changelog is at the link below, and the updated R packages are available now for download from GitHub.
RHadoop: Changelog
Comments