I am curious to know if anybody attempted to use any deep learning algorithm in this hackathon. Tree models tend to give optimal performance for this kind of problem. Kindly share your experience.
yes.. I tried ANN.. it gave 0.84438...
Wow... This is not bad. Would you mind sharing your code for the ANN approach?
I tried using fastai, but it ended in tears, I couldn't get it to beat 0.83 mark. The nn in scikit that I tried didn't come good. Since then didn't try again.
Lol... That ''...ended in tears'' part got me laughing. You must be a talented comedian.
since NN dont work for me to i stopped @0.83
Hmmn... early stopping!