A hospital in the province of Greenland has been trying to improve its care conditions by looking at historic survival of the patients. They tried looking at their data but could not identify the main factors leading to high survivals.
This is a companion discussion topic for the original entry at https://dphi.tech/practice/challenge/26
Adding an example submission list to the how to solve tab would be useful. It can just be something like this:
[0, 1, 0, 1, 1,…]
By including this, it becomes clear immediately what the expected output is.
Great challenge by the way. The dataset allows a lot of different kinds of analyses.
thanks a lot for sharing the feedback @rubenkerkhofs we shall definitely incorporate this.
Just giving heads up, we just added this as a practice challenge from our previous bootcamp and it is independent of the ongoing Deep Learning Bootcamp. Cheers!
I tried submitting my model’s prediction values in the Solve in this format:
predictions = [0.4436, 0.4554, 0.5…]
but it says wrong answer. Is something missing here?
@sunny, it expects to receive the predicted labels (not the score) i.e. ones and zeros. An example is the list [0, 1, 1, 0, 1, 0, 0, …] . You have to define a cut-off value and label everything above the cut-off as 1 and below the cut-off as 0.
I submitted my prediction, but my submission turned out to be wrong.
My prediction contains only 0 and 1, which has 9303 length.
I want to know how to submit.
Yes, @masa2197, I noticed it too! The expected submission is for some reason the training data… I don’t know why, might be a mistake
Your total number of predictions should be same as the total number of records in test dataset. Please check this.
Could you please elaborate it more what you noticed? It’s working fine from my side. I made a submission and it worked fine.
Yeah, so the evaluation dataset contains 9303 observations like mentioned by @masa2197; however, when I submitted a list of 9303 elements it rejected my submission. When I then tried to submit a list of 23097 observations (the length of the train dataset) it worked.
But tbh, I just wanted to try out the submission platform with this challenge so I did not check very thoroughly what was and was not correct
Can you tell me the length of your predicted data.
The length of predicted data should be 9330.
Hi @rubenkerkhofs @masa2197 @abhiram
Thanks for putting it across. We are looking into this issue. Once done we will let you guys know in this tread only. However there was wrong dataset link placed in test data. Our apology for this. We have changed the test dataset link which have 9330 records.
Great, thank you for letting us know!
The issue has been resolved. Please let us know if you face any.
Thanks again for letting us know!
Hi, Where can i find the test dataset?? it is not included under the data tab!
Have update it. Please check it now.
Thanks for pointing it out.