Laz4rz / CS-433-EPFL-Project1-ML

CS-433 Machine Learning - Project 1 Report

Group:

• Lorenzo Drudi - 367980
• Mikolaj Boronski - 376967
• Olena Zavertiaieva - 364500

The run.py file contains the code for the best model we obtained.

Requirements

Check and install the requirements inside requirements.txt:

pip install -r requirements.txt

Implemented Algorithms:

• Linear regression using gradient descent
• Linear regression using stochastic gradient descent
• Least squares regression using normal equation
• Ridge regression using normal equation
• Logistic regression using gradient descent
• Regularized logistic regression using gradient descent

All the algorithms are inside the implementations.py file.

Project structure

• run.py: file to run to obtain our best model.
• param_grid_search.py: run to test more parameters on the same model.
• implementations.py: implementations of all the ML algorithms.
• src/features: functions for data cleaning and feature processing (NaN replacing, downsampling, columns removal, outliers removal).
• src/model: functions for training and getting the predictions (k-fold cross-validation).
• src/evaluation: evaluation metrics (rsme, accuracy, f1).
• src/utils: utility functions (batch iteration, submission creation, weights generation).

Useful functions implemented

• Data Cleaning and Feature Processing:
  • Drop calculated features.
  • Drop rows with more than NaN% of values.
  • Remove outliers.
  • Replace NaN with mean, most frequent, or random uniform distribution between min and max.
  • Downsampling (labels balancing).
  • Standardization (z-score).
  • Polynomial expansion.
• Training process:
  • K-fold cross-validation.
• Evaluation metrics:
  • F1 score.
  • Accuracy.
  • RSME.