Project 1 of the ML course @ EPFL (CS-433)

This repo contains the machine learning system to tackle the Higgs Boson Dataset using regression techniques. THe models use only numpy and matplotlib for visualisation.

Dataset from the ATLAS Higgs Boson Machine Learning Challenge 2014 - see here

• /EDA contains notebooks for data exploration and feature engineering exploration
• /dataset contains the original dataset and our test and train set.
• /helpers contains helper functions for the IO and for the implementations
• /models contains the notebooks for model and feature set exploration
• /testcontains unittests for the implementations and the helpers
• implementations.py contains the implementations of the 6 machine learning algorithms and some extension/additions to these algorithms
• preprocessing.py contains the preprocessing pipelines that we use on the raw dataset before training
• run.py is a script that reproduces our top model predictions, it creates a submission.csv file that contains the predictions
• plotting_regression_*.py are scripts for plotting the behaviour of the logistic regression for different learning rates, these were used to gain insight into the influence of the learning rates

Documentation according to the docstring - reST style

see https://sphinx-rtd-tutorial.readthedocs.io/en/latest/docstrings.html for a small example

Testing is done using Unittest, which is the built-int test framework for python.

Tests are grouped in the /test folder. For tests to be auto-discoverable by the framework the files need to start with test and the tests need to be in a class that inherits from the unittest.TestCase baseclass.

You can then run all tests from the project root folder by entering python -m unittest in your command line.