ChaLearn Looking at People - Job Candidate Screening Coopetition Forum

Go back to competition Back to thread list Post in this thread

> Interview binary ground truth?

From the "annotation_training.pkl" we find that "interview" is a regression problem; However, we are asked to give a binary decision without having a threshold. Is there something missing or are we acquired to find the solution. Also, without ground truth of "interview" (binary descition Yes/No) we can't calculate neither the accuracy nor (FP, FN and EER).
Could you explain this problem (missing the binary ground truth of the interview) and how to address the issue (classification or regression).

Posted by: Bekhouche @ Jan. 16, 2017, 8:09 p.m.

Dear Bekhouche
Thank you for commenting this. Exactly, it is a REGRESSION problem on the target feature. Sorry for the misleading information regarding "binary classification". The information will be correctly updated in few hours.
best regards

Posted by: sergio.escalera.guerrero @ Jan. 17, 2017, 6:59 a.m.
Post in this thread