Primary competition visual

bloods.ai Blood Spectroscopy Classification Challenge

Helping Global
$7 500 USD
Completed (~4 years ago)
Classification
1103 joined
265 active
Starti
Oct 19, 21
Closei
Feb 13, 22
Reveali
Feb 13, 22
Updated starter notebook
Notebooks · 4 Feb 2022, 16:55 · 2

https://github.com/suman560/bloods.ai-Blood-Spectroscopy-Classification-Challenge/tree/main

Best baseline score was obtained using support vector classifier with accuracy 0.7027

Discussion 2 answers

Hi Suman-123,

Thank you for posting this, I was just having a look at the notebook and I see that the score for multi_target_forest.score(features, labels_n) is 0.226 on the training set. Do these trained SVM model give a 0.7027 on the test set? I ran the code myself aswell and found that the accuracy for each respective label is roughly 0.5 : 0.8: 0.5.

Im struggling to understand how this gets 0.7027 on the test set. I might be missing somthing, could you perhaps help me understand this and point me in the right direction?

4 Feb 2022, 20:08
Upvotes 0

Hy DylanGeldenhuys,

I do not really have the understanding of the data.In order to create training data i simpy averaged the data from 60 readings of same donation id. After that i simply applied some machine learning model simultaneously for all three output as given in starter notebook.

Among all the model i used ,SVM achieved best over all accuracy on public test set.