sparksummiteast
Time Series Talk intro: why we do it & "the whos who of who cares" - Review the RDD * distributed, fault tolerant, in memory computations * show the type and the operations supported by the type - Pair RDD * show the type and operations e.g. cogroups - introduce time & time series (definition) * time as a group e.g. partitioning by time w/ Pair RDD * why is this inefficient - Ordered Pair RDD * use ordering in partition for efficient operations at scale * what operations can we do with an ordered Pair RDD (joins on order) - Time Series RDD * as-of joins * window functions - Time Series RDD in Scala * yes, but ... users want python * and Pandas * which is fine. but toPandas #%%%^$^ - Time Series in Python * examples * performance benchmarks - Looking Ahead * toPandas + Arrow * Catalyst work in DF