Project 1 of the CS433-Machine Learning course given at the EPFL Fall 2021.
- Camillo Nicolò De Sabbata (@cndesabbata)
- Gianluca Radi (@radigianluca)
- Thomas Berkane (@tberkane)
The goal of this project is to find the machine learning classification model that would best predict the Higgs boson decay signatures from background noise. Our best result was achieved using the Ridge regression method, with a categorical accuracy of 0.840 and an F1 score of 0.756 on AIcrowd.
implementations.py
: contains implementations of Least Squares Regression (Normal, with GD/SGD), Ridge Regression (Normal) and (Regularized) Logistic Regression with GDrun.py
: main executable to recreate our best score on AICrowdhelpers.py
: some helper functions used by different modulesdata
: contains the datasets (.gitignore'd)README.md
: this file
Python modules requirements: numpy
, matplotlib.pyplot
, typing
and csv
. Predictions will be saved in the data
folder. To reproduce our best score with ridge regression that we submitted on AIcrowd:
python run.py