This is our project for the Machine Learning Course at EPFL - 2019. In this folder you can find:
- Our report
- Notebooks: Folder containing Data analysis notebook and implementation of exercise 10 for ALS algorithm
- BlendModels: Notebook representing our best model - Its python version is run.py
- als.py: Functions necessary to implement ALS algorithm for the best model
- helpers.py: Functions needed for ALS algorithm. This folder comes from exercise 10 solutions
- implementations.py: Functions required for best model. Very useful for refactoring the code inside notebooks
- run.py: Python file of our best model
- validation_gridsearch: This notebook computes the optimal weights of each model (expanded with feature expansion) It does a grid search for each algorithm individually, train them on a train set, and take predictions on a validation set. It then run a ridge regression using Scikit to obtain optimal weights that we copy inside Blendmodels
If you don't have surprise, you can run run.py using the following command: ipython run.py.
First line of the code will install surprise for you.
Please make sure to update the path for data set and sample submission csv files.