Benchmark

benchmarking dplyr filter versus base subsetting methods