Getting missing data points percentage on a timeseries


#1

We have an important number of sources > 1000, and several timeseries to look at (tens).
What is the fastest way to get the percentage of missing data in a given period for each of those timeseries.
(using evalMetrics may take several minutes)


#2

This seems pretty embarrassingly parallel, so a MapReduce would be easy to write to speedup your computation.