bellonet / docker-demo-jupyter-mongo Goto Github PK

This repo contains a docker-compose (multi-container) application to facilitate the classification of ECG time-series data.

The components of the tool:

1. A docker-compose with two contrainers:
   • Docker_JL - uses an image based on jupyter/tensorflow-notebook:
     https://hub.docker.com/r/jupyter/tensorflow-notebook
   • Docker_DB - uses the mongoDB docker image:
     https://hub.docker.com/_/mongo
     and loads to the database the following ECG dataset:
     https://zenodo.org/record/5711347#.YxuO4NVBzJW
2. A demo jupyter notebook for the use of the multi-container docker in the classification of the patients' time series to their diagnosis:
   • A short demo of access to the database.
   • A demo of a small and simple tensrorflow network to classify the ECG data.

Docker and docker-compose must by installed.
ubuntu instructions:
https://docs.docker.com/engine/install/ubuntu/
https://www.digitalocean.com/community/tutorials/how-to-install-and-use-docker-compose-on-ubuntu-20-04

1. Open a terminal window.
2. clone this repo - git clone https://github.com/bellonet/docker-demo-jupyter-mongo.git
3. cd docker-demo-jupyter-mongo to have it as the working directory.
4. Inside the directory - Run docker-compose up -d
   • As this is the first time that you run the container, it would download and build all the components, thus this step would take some minutes.
   • up creates and starts the containers.
   • d detached mode: Run containers in the background. You can also run it without d but then ctrl+C would kill the process (and all its containers).

Yay - your docker is running and ready to run your analysis.

5. Open a web browser.

6. Type in localhost:8888/ in the address bar and click enter.

7. Enter daki in the token field.

8. From here you can start a new jupyter notebook and perform your analysis. Alternatively, you can open the demo_ecg.ipynb to learn a bit how to use mongo from python (using pymongo), learn how to use the data in the ECG database, and how to run a simple tensorflow based classifier.

   • Importantly, all created notebooks to be used with this docker should be in the project home directory.

9. Once done, you can stop and remove the containers with docker-compose down.

10. You can restart the containers again with docker-compose up -d.

• While docker is running you can access a shell from each container using those commands:
  • docker exec -it Docker_JL bash
  • docker exec -it Docker_DB bash
    • From here you can enter the mongo shell by typing mongosh.

I used the following resources:
https://github.com/jupyter/docker-stacks
https://github.com/jupyter/docker-stacks/tree/main/tensorflow-notebook
https://dagshub.com/blog/setting-up-data-science-workspace-with-docker/#setting-up-your-data-science-docker-container-a-step-by-step-guide
https://faun.pub/initialize-mongodb-running-on-a-docker-container-889a43c5668a