Refinements to code efficiency

[MichaelChirico]

• GDP_dollar already read as numeric by fread. If this may not be the case, better way to handle this is to include colClasses = list(numeric = 'GDP_dollar').
• latest_date unused in this script
• index in fread is a recent feature I believe. Can revert if prefer a less bleeding-edge dependency.
• round(.SD) was 1/2 as fast as lapply(.SD, round) for me
• if not using any of the other options to predict.lm, can just access the stored fitted.values directly from the lm object
• iso_a3 %in% regression_countries in line 50 is redundant to the same in line 45. If there for emphasis, can use intersect(regression_countries, big_mac_countries) to skip the potential double vector scan

[futuraprime]

Thanks, Michael—we'll incorporate these.