The repository contains the code and report for Machine Learning course 2021 (CS-433) project 2 at EPFL. The purpose of this project is to classify snowflakes with hydrometeors class and estimate their riming degree.
The project is accomplished by team Ma_BG with members:
- Marie-Alix Gillyboeuf: [@GILLYBOEUF]
- Baptiste Hernette: [@Bapitou]
- Gaspard Villa: [@gaspardvilla]
The data for both hydrometeor classification and riming degree are on a google drive, it is not necessary to download them as the file ‘dataloader.py’ do it directly.
The project has been developed and test with python3.6.
The required library for running the models and training is numpy, panda and sklearn.
The library for visualization is matplotlib.
Results to predict the test datasets are generated by running:
run.py.
models.py: the implementation of the 4 machine learning models to train with their hyperparameters to tuned.
run.py: the results obtained with the models for both hydrometeor classification and estimation of riming degree.
tutorial.py: a small run to show how the cross validation for hyperparameters tuning was implemented.
dataloader.py: permits to load the data from a google drive and obtain the classes to estimate.
Dataprocess.py: process the data by taking the columns we need and processing standardization.
helpers.py: useful tools data preprocessing and to load our save the different models.
cross_validalidation.py: using cross-validation to find the best hyperparameters with the test accuracy of different models.
plots.py: functions that gives us the tools to visualize the data.
ML_Snowflakes.pdf: a 4-pages report of the complete solution.