This repository is my first real implication on a kaggle competition. All others end up to dropped notebooks and it has been 3 months i don't touch pandas and so on...
Here i want to make a robust pipeline as a test for later competitions. It has to train a model and submit data. Moreover it has to make continuous training test and each version of the model has to be saved for later inspection.
First, look at the jupyter notebooks which is a draft of my final model, then go into code.
I would like to set up pipelines which will structure the model because i figured out that it was a real challenge back in industry.
I will certainly use Pandas as a basic package for data exploration and sklearn. I also would like to be introduced to JAX or go deeper to tensorflow probability.