'NeMozeNamNikoNista' team repository used for managing and submitting Project 1 of the EPFL's Machine Learning course (CS-433).
Team members:
- Dino Mujkić (dinomujki)
- Hrvoje Bušić (hrvojebusic)
- Sebastijan Stevanović (sebastijan94)
- Python 3 - installation and instructions
- NumPy - installation and instructions
To get started you must obtain the data used for the EPFL's ML Course - Project 1 competition available on Kaggle.
Following files are required:
train.csv
test.csv
Data needs to be placed in the existing ./data
subdirectory.
Running run.py
from the ./regression
subdirectory will generate submission file submission.csv
in the ./data
subdirectory.
Files located in the ./data_wrangling
subdirectory contain functions used for initial data loading and manipulation, as well as the creation of the final submission files. The methods applied in the initial data manipulation phase are explained in the project's report.
Files connected to regression implementation can be found in the ./regression
subdirectory.
cross_validation.py
- functions used for cross validation testingimplementations.py
- the file contains 6 necessary algorithm implementations for this project as well as appropriate helper functionspolynomial.py
- a function used to construct polynomial basesrun.py
- a main execution file which performs data parsing and model construction to generate the optimal submission resultstest_run_implementations.py
- function used to test the implementations using a dummy dataset
Files connected to neural network implementation can be found in the ./neural_network
subdirectory.
data_loader.py
- utility functions for working with.csv
files when loading test and validation sets, and storing resultsgenetic_algorithm.py
- implementation of generational genetic algorithmnetwork_layers.py
- abstraction of network layer used by a neural network with several implementationsneural_net.py
- a high level object that consists of network layers, and can for a given vector of input instances return the vector of outputs for all the input instancesrunner.py
- execution file which generates predictions using neural network modeltransfer_functions.py
- transfer functions that can be used on the neuron's output
We encourage you to fork or clone this repository for future development. Issue reporting and pull requests are welcome as well.