Project: Train a model and deploy as an API via FastAPI on Heroku

(This project is a part of my Udacity Nanodegree course project: The skeleton for the project is in this repo. :
https://github.com/udacity/nd0821-c3-starter-code)

• This project has CI / CD enabled via GitHub Actions and Heroku

Environment: python 3.9

FastAPI

Local API

• Start and run FastAPI app and send get/post requests: main app script: starter/main.py

cd starter
python uvicorn main:app --reload

To send a POST request open another command window: starter/infer_post_request.py

cd starter
python infer_post_request.py

Live API

• Note: before sending requests to Live API, make sure the heroku app is deployed and is live
  https://census-data-app1.herokuapp.com/

To send a POST request to live API: infer_live_post_request.py

python infer_live_post_request.py 

Training and Data slicing:

• starter/starter/train_model.py
• starter/starter/train_model_on_slices.py

cd starter
python starter/train_model.py
python starter/train_model_on_slices.py 

Data Preprocessing and Model implementation

• starter/starter/ml/data.py
• starter/starter/ml/model.py

Dataset, saved models, logs, screenshots:

• Dataset: starter/data/
• Model: starter/model/
• Screenshots: starter/screenshots/
• Logs: starter/logs/slice_output.txt

Tests:

• API test script: starter/test_main.py
• Model test script: starter/starter/ml/test_model.py

cd starter
pytest