Experiments with docker-compose

docker-compose.yaml

version: '3'

services:
  client:
    #image: ubuntu:18.10
    # build : This directive can be used instead of image. Specifies the location of the Dockerfile that will be used to build this container
    build: ./
    container_name: Chat_client
    volumes:
      - ./:/code
    ports:
      - 5001:5000
    environment:
      DEBUG: "true"


  server:
    build: ./
    #image: ubuntu:18.10
    container_name: Chat_server
    volumes:
      - ./:/code
    ports:
      - 5002:5000
    environment:
      DEBUG: "true"

flask_db.py

from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.naive_bayes import GaussianNB
from sklearn.metrics import accuracy_score
from sklearn.externals import joblib

from mongoengine import *
from datetime import datetime
from flask import Flask, request
from flask_restful import Resource, Api

APP = Flask(__name__)
API = Api(APP)
connect('mlaas', host='localhost', port=27017)
try:
    pickled_model = joblib.load('model.pkl')
except:
    pickled_model = None

class TrainCollection(Document):
    time = DateTimeField(default=datetime.now(), assign=True)
    params = DictField(required=True)

class PredictCollection(Document):
    time = DateTimeField(default=datetime.now(), assign=True)
    features = DictField(required=True)

class Train(Resource):
  def post(self):
    model_kwargs = request.get_json(force=True)
    if len(model_kwargs) > 1:
        return "Invalid Parameters"
    collection_obj = TrainCollection()
    collection_obj.params = model_kwargs
    collection_obj.save()
    if model_kwargs.get('priors') == 'None':
        model_kwargs['priors'] = None
    data = load_breast_cancer()
    labels = data['target']
    features = data['data']
    train, test, train_labels, test_labels = train_test_split(features, labels, test_size=0.33, random_state=42)
    print(test.shape)
    gnb = GaussianNB(**model_kwargs)
    model = gnb.fit(train, train_labels)
    joblib.dump(model, 'model.pkl')
    preds = gnb.predict(test)
    #print(preds)
    accuracy = accuracy_score(test_labels, preds)
    return {"Accuracy Score ":accuracy}

class Predict(Resource):
  def get(self):
      for predictions in PredictCollection.objects:
          print(predictions.time)

  def post(self):
    global pickled_model
    features = request.get_json(force=True)
    collection_obj = PredictCollection()
    collection_obj.features = features
    collection_obj.save()
    if pickled_model == None:
        try:
            pickled_model = joblib.load('model.pkl')
        except:
            return "Please Train your model"
    else:
        model_ = pickled_model
    prediction_result = model_.predict(features['features'])
    print(list(prediction_result))
    return {"Prediction Results": str(prediction_result)}

API.add_resource(Train, '/train')
API.add_resource(Predict, '/predict')
APP.run(host='0.0.0.0', port=5000, debug=True)