Closed russointroitoa closed 4 years ago
Hey,
That looks like a bug in the dataset generator. I'll investigate it when I get time next week, thanks for reporting it.
It is a bug in the dataset. I'll be updating the dataset with the correct values sometime next week.
Sorry, I think that the error is still present in the second version of the dataset.
@russointroitoa Isn't that due to the half time switch, or am I missing something?
Ok, I didn't realize the half time switch, now I understand the logic! Thanks!
Dear Organizers,
Exploring the dataset, I noticed an inversion of the meaning of _currentstate position values: in the description of the data format, it is explained that the first position of _currentstate refers to CT and the second position to Terrorist. Considering dftrain in the notebook provided, the first round goes from index 0 to 3 where the _roundwinner = CT, such that in the next round (starting from index 4) the victory point needs to be assigned to CT; however, the point is assigned to the second position of _currentscore, that is Terrorist. Moreover, there is another inversion in the meaning of _currentstate: the round defined from row index 76 to 79 is won by CT and the second position in _currentstate is incremented, following the same approach so far. The next round (from row index 80 to 83) is won by Terrorist, but always the second position of _currentstate is incremented.
I would like to know if I'm missing something or this is a real issue.