This code is our solution to EPFL's CS-433 Machine Learning course competition : EPFL Machine Learning Higgs
Our implementation achieves an accuracy of 81.5%.
proj1_helpers.py
: file provided by the teaching team.implementations.py
: contains our implementations of the labs' functionscross_validation.py
: contains our cross validation codecross_validation_run.py
: performs cross validation to produce the best hyper-parameters for our model (stored inbest_degrees.npy
,best_lambdas.npy
)data_manipulation.py
: contains our helper methods to perform data manipulation and feature engineeringrun.py
: contains the code we use to train our model and make our predictions (needsbest_degrees.npy
,best_lambdas.npy
)best_degrees.npy
: contains the best degrees for feature augmentation. Can be reproduced by running the scriptcross_validation_run.py
best_lambdas.npy
: contains the best lambda for our machine learning model training. Can be reproduced by running the scriptcross_validation_run.py
submission.csv
: tabular file containing our predictions. Used for the submission on the challenge website
To execute our code, the dataset must be downloaded from here. The .csv
must be extracted from the archive and placed in a folder called data
placed on the root of the project.
In case any help is needed:
- Ahmed Jellouli : ahmed.jellouli@epfl.ch
- Ayman Mezghani : ayman.mezghani@epfl.ch
- Karim Hadidane : karim.hadidane@epfl.ch