Today we were made aware of some inconveniences in the Perl script for splitting your training and test set from the public challenge data:
- the output file names were not consistent with the description (fixed)
- the anonymous user ids were anonymized another time (fixed)
- different date representation was expected (fixed)
The script is still not user friendly, but at least it prints out some messages now. We added a FAQ entry which exemplifies the process of splitting the public data.