Notrix Web

Notrix Web serves to visualize relationships between solutions to kaggle competitions. Also you can upload your own solution to a competition or whole solution sets for the application to analyze.

The application is amde of two distinct parts:

• Notrix-Web: A Vaadin application that is the frontend and backend to the web application
• Server: A python FastAPI server that processes uploaded competitions and notebooks

Starting in Development Mode

To start in development mode you need a postgres database. This is probably easiest to start inside a docker container with its port exposed, or as a service on your local machine. Once you have that database, you can adjust the database parameters in the application-dev.properties file.

After this you can start the application using ./mvnw from inside the java directory. Alternatively you can start the application in debug mode from inside IntelliJ (which I recommend for development), as this gives you the ability to hot-reaload the application.

Lastly you have to start the FastAPI server. For this, first create a python environment of your choice (I recommend using venv like python3 -m venv .venv) inside the python directory. After activating the environment (source .venv/bin/activate) install the dependencies using pip install -r requirements.txt. This might take a while, as the required packages are wuite large. Once everything is successfully installed, you can run the FastAPI server in development mode for hot-reloading using fastapi dev app/server.py. This will host the application on port 8000 by default. If you change this, make sure to also change the application-dev.properties. Make sure to either set the "OPENAI_API_KEY" environment variable or put a .env file in the app directory for the GPT analysis to work.

Building for Production

To deploy this application, you can simply use the provided docker-compose.yml file. Make sure to add a .env file in the top directory with your environment variables or replace the placeholders inside the compose file.

Initializing the database

Once your database and the application are running, the application will automatically create all required tables, relations and constraints. To actually get data into the database, you can copy the data from csv files using the commands inside sql/import.sql. To connect to the databse in the first place, you can use psql -U postgres -d kaggle-vis --password. If you're running the database inside a docker container, prepend docker exec -ti <container-name> to the command. In that case also make sure to pass the generated data files into the container using volumes. All the data can be generated from meta-kaggle and meta-kaggle-code using the scripts in the python directory.