server.py

import flask
import pandas as pd
import os
from sklearn.neighbors import NearestNeighbors
import sys
from joblib import load

# load the model
clf = load('my_model.joblib')

# define a predict function as an endpoint
def make_prediction(dictOfInputs):
    x=pd.DataFrame.from_dict(dictOfInputs, orient='index').transpose()

    isKneighbor = isinstance(clf, NearestNeighbors)

    if isKneighbor:
        return clf.kneighbors(x)[0].tolist()

    return clf.predict(x)[0].tolist()

@app.route("/", methods=["GET","POST"])
def predict():
    data = {"success": False}

    body = flask.request.json
    params = None

    if (body == None):
        body = flask.request.args

    if "inputs" not in body.keys():
        data["message"] = "The inputs parameter is missing. Try and use {inputs: [array of inputs to the model]}"
    else:
        params = body["inputs"]

    # if parameters are found, return a prediction
    if (params != None):
        dictOfInputs = { i : params[i] for i in range(0, len(params) ) }
        try:
            data["outputs"] = make_prediction(dictOfInputs)
            data["success"] = True
        except Exception as e:
            data["message"] = "There was an error while running your model: " + str(e)

    # return a response in json format
    return flask.jsonify(data)

# start the flask app, allow remote connections
app.run(host='0.0.0.0')