wonyoungseo / ex-census-bureau-classification-deployment

Census Classification ML API

Income category prediction project based on Census data (Udacity Nanodegree project)

Prediction API is currently running on following URI

• https://census-classification-udacity.herokuapp.com/

Key takaways from this project

• Data EDA (jupyter notebook)
• ML model training (scikit-learn)
  • feature engineering
  • model training
  • ML model performance monitoring on sliced data
• Unit test (with pytest)
• Version Control
  • code (git, Github)
  • data (git, DVC)
  • model (git, DVC)
• Remote storage (AWS S3)
• Continuous Integration (Github Actions)
• API development (FastAPI)
• Continuous Deployment (Heroku)

Sending reqest to API

Via Python script (refer to example_api_post_heroku.py)

import json
import requests

headers = {
    "accept": "application/json",
    "Content-Type": "application/json",
}

url = "https://census-classification-udacity.herokuapp.com/prediction"

data = {"age": 40,
        "workclass": "Private",
        "education": "Doctorate",
        "marital-status": "Married-civ-spouse",
        "occupation": "Prof-specialty",
        "relationship": "Husband",
        "race": "White",
        "sex": "Male",
        "native-country": "United-States",
        "hours-per-week": 60}

response = requests.post(url, headers=headers, data=json.dumps(data))
print(response.status_code)
print(response.json())