Skip to content

Tutorial accompanying the youtube video .. meant for educational purposes. The Machine Learning done in this project is very simplistic and the repo is meant to shortly show how to build a ML API.

Notifications You must be signed in to change notification settings

SandroLuck/TutorialMachineLearningAPI

Repository files navigation

General

This is the repository associated with the youtube tutorial at: TODO: It is very simplistic and is meant as a guideline for implementing a Machine Learning API and not necessarily a good Machine Learning algorithm. I put particular emphasis on making this tutorial as short as possible.

To run:

Locally

I recommend using docker below, but also works like this: pip install -r requirements.txt cd TO/PATH uvicorn main:app --reload

Using Docker

cd to the folder: cd PATH/TO/MlApiTutorial run docker: docker-compose up --build

Find Solution at

recommended: go in browser to http://localhost/docs (Assuming your docker hosts there else check docker ps it's on port 80) you can test the API directly with GUI http://localhost/docs or sent to http://localhost/will_survive

For dev purposes

tests: python -m main run server: uvicorn main:app --reload

Troubleshoot

find IP: docker ps

Files

'main.py' holds the API-related code. We put all in one file to make project simpler. 'test.py' has one test case to validate classifier is roughly working. 'Dockerfile' configures the server and installs dependencies. I used the Uvicorn/FastApi 'docker-compose.yml' not really needed. Simplifies execution on your side. I put restart: always incase you test edge cases (I did not handle them). 'make_model.py' builds and selects a model. I did not validate them much since this is more of an API tutorial.

Test

I implemented a test to integrate it. In practice running them in Travis, Circelci before pulling from GitHub would be better.

Deploy to a cloud

Check docker related deployment possibilities of your favorite cloud provider

Data From

Kaggle: https://www.kaggle.com/c/titanic-dataset/data

About

Tutorial accompanying the youtube video .. meant for educational purposes. The Machine Learning done in this project is very simplistic and the repo is meant to shortly show how to build a ML API.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published