Smoke test marked as failed despite correctly formatted submission.csv file is generated

DaT Parkinson's Prediction Challenge
1

Hello!

I am submitting my code for smoke testing and it is marked as “failed”. The log shows that my script exited without errors. I added a debug print to check the content of my final submission.csv file and it looks healthy:

At the end of main.py:

with open('/code_execution/submission.csv' 'r') as file:`
    file_content = file.read()`
    print(file_content)`

This gave in the ouput:

uid,is_pathologic
v23gxj0m,0.6161804105900507
1r11q7uw,0.044185568153099174
cwbfuk9e,0.6816717708095468
t7zewviy,0.6011349427804435
w69hzgop,0.9367508045701831
(etc)

This confirms that:

  1. the submission.csv was correctly created, at the location described in the docs
  2. the content is compatible with the expected format (header, coma-separated) etc

I fail to see why such predictions would not be compatible with a score evaluation.
Or it is because smoke tests don’t get scored?

Thanks for your help

arovai

2

Hi @arovai , it looks like the ordering of your UIDs is incorrect - they should be in the exact same order as the submission_format.csv in the smoke test execution environment, which you can read from /code_execution/data/submission_format.csv. Hope this helps!