This dataset has the following directory structure: year_month=?/timestamp_sod=?/?.csv.gz
 - year_month indicates the year and the month which the CSVs in that directory cover.
 - timestamp_sod indicates the day which the CSVs in that directory cover (UNIX timestamp at the Start Of the Day).
 - the precise filename has no particular meaning.

Each individual CSV contains a header row. In total the dataset has approximately 3.5 billion rows.