Simple web application that allows you to run and track a simple, real-time, kaggle-like ML challenge.
To run the project you just need to execute the command below:
docker-composer up
The datasets need to be placed inside the folder 'datasets' where each one must have 2 files. NAME.x.npy and NAME.y.npy. Please check the file temp/create-dataset.py to learn how to create a new dataset.
The uploads folder will be used to store temporarily the models submitted.
The only way to delete submitted results is by manually editing the "database" file 'super-secure-database.json'. Have fun with that ;)
Students need to upload the models serialized using the library joblib. please check the file temp/create-model.py to see how the students need to create the models:
from joblib import dump
dump(clf_lr, 'model.joblib')
To upload the model the student just need to use the "Submit Model" form: