Epistemic Conformal Score (EPICSCORE).

Here we provide an implementation of EPICSCORE and notebooks to reproduce the figures and tables from the paper Epistemic Uncertainty in Conformal Scores: A Unified Approach (check paper here).

We demonstrate how to use our method on regression and quantile regression base models in two demo notebooks.

Installing Dependencies and Package

To install the necessary dependencies and the EPICSCORE package, follow these steps:

1. Navigate to the directory containing the setup.py file.

2. Activate conda in the terminal

   source activate

3. Install the local conda environment with all dependencies by running the following command:

   conda env create -f EPICSCORE_env.yml

4. Activate the conda environment:

   conda activate EPICSCORE_env

5. Install the EPICSCORE package:

   pip install .

Alternatively, if you use Poetry, follow these steps:

1. Navigate to the directory containing the setup.py file.

2. Initialize a new Poetry environment:

   poetry init

3. Add the required dependencies to the Poetry environment:

   xargs poetry add < requirements.txt

Running Real Data Experiments

To download and process all required data, execute the following command:

bash data/data_scripts/download_data.sh

To run all experiments for quantile regression, use the command:

python Experiments_code/metrics_real_data.py