README.md

Coffee Quality prediction app ☕

This code runs a small flask based webserver which can make predictions on the given coffee samples total cupping score provided the following attributes are passed as a JSON.

Training the model

The script train.py can be used to train a model with the final model and parameters. After running it will print the current model and parameters as well as the auc and rsme scores for the model. A model.bin artifact will be saved in the ./models directory. Note that these are ignored by git and therefore not saved to github. You need to run this before you can run the application.

Running locally 🖥️

Install dependencies ⚙️

First set up the virtual environment as documented in the root README. Then run:

pipenv install

Running the application ▶️

The below command will run the application locally within the virtual environment.

make dev

Running with Docker 🐳

The below command will both build the Docker image and run it.

make run-with-docker

Example call to the service

Using Curl

curl --header "Content-Type: application/json" \
    --request POST \
    http://localhost:8000/predict \
    -d @test_data/coffee_sample.json

Using Python

import requests

coffee_sample = {
    "moisture": "0.12",
    ...
}

url = "http://localhost:8000/predict"

requests.post(url, json=coffee_sample).json()
# returns python dictionary of response

There is a sample code for this here which you can run with pipenv run python request.py.

Deployment with Heroku

Run the following commands to deploy this project to Heroku.

# login into Heroku
heroku login
heroku container:login

# only need to run this the first time to create the app
make create-app

make push-app

make release-app
``