A simple example of how Django can be used to serve predictions through an API.

Open up a terminal in the repositories root and run these commands to align your environment:

conda create -n titanic-django

conda activate titanic-django

conda install pip

pip install -r requirements.txt

This will host the API.

• Open a terminal and navigate to the titanicapi project folder (first level, not the second). Then run:

conda activate titanic-django

(making sure you are executing from within the new environment)

Next run

python manage.py runserver

This will run a local development server of the django app. The model's API is now ready to call!

• Run demo.ipynb in the titanic-django conda environment once more (or any environment with requests and pandas installed)

• You should be getting results from your server!

That's it! Congrats on your first Django model API!

• Run the following in your desired project directory:

django-admin startproject titanicapi

python manage.py migrate

django-admin startapp api

• Add api to installed apps in main settings

• Add two files to api app directory:

1. functions.py

• here you write your functions for loading the serialized model and classifying passengers

2. urls.py

• here you specify the URL on which your API will sit. This will allow you to send post requests with feature data to get predictions.

• Create APIView class in view.py with a post function to handle post requests. E.g.

get_classification(APIView)

• Add api urls to project root urls (titanicapi/urls.py) so that the main app knows that the api urls exist. E.g.

path('api/', include('api.urls'))

• Try running your server to see if it likes your setup. Generally Django logs are very good.

python manage.py runserver