Skip to content

Latest commit

 

History

History
57 lines (43 loc) · 2.25 KB

README.md

File metadata and controls

57 lines (43 loc) · 2.25 KB
Raw

Project 1 : Higgs Boson Challenge

This is a project realised for the Machine Learning Course [CS-433]. We were working on finding the best model for the Higgs Boson Challenge. After testing 6 models we ended up using Logistic Regression for our final submission.

Folder Structure


project1
├── data                               # The train and test data sets
│   ├── test.csv.zip
│   └── train.csv.zip
├── results                            # Final results for submission
│   └── results_final.csv
├── Report.pdf                         # The report
├── README.md                          # The ReadMe file
├── run.py                             # Script for training the models with the train set and get predictions for the test set
├── scripts                            # Scripts used for testing the models and analysis of the data
│   ├── hyperparameterTestResult.txt   # Results for the hyperparameters after crossvalidation
│   ├── dataAnalysisPlot.py
│   ├── testing_parameters.py
│   └── Test_Models.ipynb
└── utils                              # The functions used in the scripts
    ├── crossvalidation.py             # Functions for performing crossvalidation
    ├── features.py                    # Dictionaries containing infromations for the columns
    ├── helpers.py                     # Helping functions
    ├── implementations.py             # Mandatory functions
    ├── loss_gradient.py               # Functions for the computing the loss and the gradients
    └── preprocessing.py               # Functions for feature preprocessing

Zipped Files

Before running the code you will need to unzip the files containing the train and test data:

  • data/train.csv.zip
  • data/test.csv.zip

Running the code

To run our code and get the final predictions you can use the following command:

python3 run.py

Final submission results

The final results after running the run.py script can be found in the results folder under the name results.csv

AICrowd

Our final results on AIcrowd gave a 0.814 accuracy .

Authors

  • Marija Lazaroska
  • Deborah Scherrer Ma
  • Méline Zhao