main.py

from fastapi import FastAPI
from joblib import load
import json

# FastAPI is a framework for creating web APIs
app = FastAPI()

'''
GET Request
Path structure: http://localhost:8000/

A standard controller to confirm you are able to connect to your API.
'''
@app.get("/")
def index():
    return "This is a REST API for enabling my ML model to be used by other services."

'''
POST Request
Path structure: http://localhost:8000/predict?color={enter_color}&shape={enter_shape}

A controller for predicting a base happiness score for a pokemon based on their color and shape.

Valid values for query parameters:
color: ["black", "blue", "brown", "gray", "green", "pink", "purple", "red", "white", "yellow"]

shape: ["armor", "arms", "ball", "blob", "bug-wings", "fish", "head", "humanoid", "legs", 
"quadruped", "squiggle", "tentacles", "upright", "wings"]
'''
@app.post("/predict")
def predict_happiness(color = 'black', shape = 'armor'):

    # Extract dictionaries saved from our jupyter notebook that map a category code to its name.
    with open('static/mappers/color_mapper.json', 'r') as colorfile:
        color_mapper = json.load(colorfile)
    with open('static/mappers/shape_mapper.json', 'r') as shapefile:
        shape_mapper = json.load(shapefile)

    # Map the query parameters "color" and "shape" to its corresponding category codes
    color_code = [int(k) for k,v in color_mapper.items() if v == color][0]
    shape_code = [int(k) for k, v in shape_mapper.items() if v == shape][0]

    # Load the model we saved from our jupyter notebook and predict a happiness score.
    model = load('./static/my_model.joblib')
    score = model.predict([[color_code, shape_code]])

    return { "happiness_score": score[0][0] }