https://github.com/arnabsharma91/deepSSTA.git

Repository implementing the approach of team UPB-DICE

Installation

• Clone this repository: git clone https://github.com/arnabsharma91/deepSSTA.git
• Create a conda environment named ecmldeep: conda create -n ecmldeep python=3.10.14
• Activate the environment: conda activate ecmldeep
• Install all dependencies: pip install -r requirements.txt
• If working on Jupyter Lab or Jupyter Notebook, also install the kernel: conda install ipykernel && python -m ipykernel install --user --name ecmldeep --display-name "ecmldeep"

Training our models:

• Run the notebooks Train_LightGBM_CatBoost.ipynb, Train_LightGBM_CatBoost_more.ipynb, NN.ipynb, and NN_more.ipynb

Predicting for the Development Phase

• Run the notebook Predict_Development_Phase.ipynb

Predicting for the Testing Phase

• Run the notebook Predict_Test_Phase.ipynb

Predicting for September 2024

• Run the notebook Predict_September2024.ipynb. Predictions for September 2024 are available both in the technical report and in this notebook.

Remark. Our scores for both the development phase (0.09913) and testing phase (0.0872) are clearly visible in the notebooks Predict_Development_Phase.ipynb and Predict_Test_Phase.ipynb.