When creating the two data frames in the function, we extract a subset of values from each record, create a matrix and then a data frame. This essentially involves creating at least 3 copies of the relevant values in memory. What are they? Are there better ways to do this? Consider the time, x, y data frame, for example. We could use l apply (), rather than sapply(), to return a list of vectors (say, list Of Vectors) of length 3. We could then use as data frame (do call (rbind, list Of Vectors)) to create the data frame. Does this reduce the amount of memory used? Are there other ways to arrange the computations to avoid having 3 copies of the data in memory at once? If so, are these faster or slower than our approach above in read Log()?