In this project ECG data was classified using the ResNet. Both the JupyterLab instance and the database were created in one Docker container.

Running the project

The project is started by running run_project.bat. This will:

1. Build the docker_jl image
   Image contains all of the packages needed to run the code (~3.45 GB)
2. Run a container jlc of the docker_jl image
3. Download the ECG Rhythm LEAD 1 database and import it into MongoDB
4. Open Jupyter Lab
   The code is written Project.ipynb. It was largely sourced from https://github.com/spdrnl/ecg/blob/master/ECG.ipynb. Data is imported from MongoDB, minority classes are oversampled, the Resnet is trained and the results are tested. You can choose to train the ResNet again or load the weights with whom the best validation loss was achieved.