Last week we decided to remove the GPS records which do not have a time difference between each other greater of 1 hour.
While this choice could be motivated in many ways (The fact that users are covered by GPS for just the 2% of the time, the coarse grained of the traces ,..) lower the limit to 30mins could be useful to not miss any relevant movement.
While with 1h we ended up with 56K records, with 30mins we get 72K; which in the and correspond to 55K rows.
With these data the final dataset is so composed by 213,392 records, the time difference becomes:
Mean: 6.6833 Max: 5340.74 Min: 0.0
Number of places with a time difference over 80 hours: 2858 Number of places with a time difference less than 80 hours: 210534 Number of places with a time difference less than 20 hours: 199089 Number of places with a time difference less than 2 hours: 166958
While the median number of records per user becomes 868.0
Mean number of unique places visited by users: 45.21