Machine Learning Model Deployment with Flask

Table of Contents

  1. Overview
  2. Prerequisites
  3. Setup
  4. Creating a Flask App
  5. Importing the Machine Learning Model
  6. Defining API Endpoints
  7. Deploying the Model
  8. Testing the API
  9. Conclusion

Overview

In this tutorial, we will learn how to deploy a machine learning model using Flask, a popular Python web framework. Flask allows us to create a RESTful API that can be used to interact with our machine learning model. By the end of this tutorial, you will be able to create a Flask application, load a pre-trained machine learning model, define API endpoints, and deploy the model for production use.

Prerequisites

Before starting this tutorial, you should have a basic understanding of Python programming and machine learning concepts. It is also recommended to have the following software installed on your system:

  • Python 3
  • Flask
  • scikit-learn (for the machine learning model)

Setup

To get started, let’s set up our development environment. Follow these steps:

  1. Install Python 3 from the official website: https://www.python.org/downloads/
  2. Open a terminal or command prompt and verify the installation by running the following command: 
     python --version
  3. Install Flask by running the following command: 
     pip install flask
  4. Install scikit-learn, a popular machine learning library, using the following command: 
     pip install scikit-learn

    Creating a Flask App

Now that the setup is complete, let’s create a Flask application. Follow the steps below:

  1. Open a text editor or an integrated development environment (IDE) of your choice.
  2. Create a new Python file and name it app.py.
  3. Import the necessary Flask modules: 
     from flask import Flask
  4. Create a Flask app instance: 
     app = Flask(__name__)
  5. Add a basic route to the app: 
     @app.route("/")
 def hello():
     return "Hello, World!"
  6. Run the Flask app. In the terminal, navigate to the directory where the app.py file is located and run the following command: 
     python app.py
  7. Open a web browser and visit http://localhost:5000. You should see the message “Hello, World!” displayed.

Congratulations! You have successfully created a basic Flask app.

Importing the Machine Learning Model

Next, let’s import a pre-trained machine learning model into our Flask application.

  1. Create a new directory in your project called models. 
     mkdir models
  2. Place the pre-trained machine learning model file, let’s say model.pkl, inside the models directory.
  3. In app.py, import the necessary modules to load the model: 
     import pickle
  4. Load the machine learning model in the Flask app: 
     model = pickle.load(open("models/model.pkl", "rb"))

    Make sure to replace model.pkl with the actual filename of your model.

Defining API Endpoints

Now, let’s define the API endpoints that will allow us to interact with our machine learning model.

  1. Add a new route to the Flask app that accepts HTTP POST requests: 
     @app.route("/predict", methods=["POST"])
 def predict():
     # Retrieve the data from the request
     data = request.get_json()
	
     # Perform predictions using the loaded model
     predictions = model.predict(data)
	
     # Return the predictions as a JSON response
     return jsonify(predictions.tolist())

    In this example, we assume that the input data is a JSON object containing the necessary features for making predictions.

Deploying the Model

To deploy our machine learning model, we can use various cloud platforms or web hosting services. Here, we will use Heroku, a popular platform for deploying web applications.

  1. Install the Heroku CLI following the official documentation: https://devcenter.heroku.com/articles/getting-started-with-python
  2. Once installed, open a terminal or command prompt and log in to your Heroku account: 
     heroku login
  3. Create a new Heroku app: 
     heroku create <app-name>

    Replace <app-name> with a unique name for your application.

  4. Push your code to Heroku: 
     git push heroku master

    This command will deploy your application to Heroku.

  5. Scale up the Heroku dynos to ensure the application is running: 
     heroku ps:scale web=1

    You can check the status of your application using the following command:

     heroku logs --tail

    Testing the API

To test the deployed API, you can use cURL or any other API testing tool. Here’s an example using cURL: bash curl -X POST -H "Content-Type: application/json" -d '{"feature1": 5, "feature2": 10}' https://<app-name>.herokuapp.com/predict Replace <app-name> with the name of your Heroku app.

If everything is set up correctly, you should receive a JSON response containing the predictions made by your machine learning model.

Conclusion

In this tutorial, we learned how to deploy a machine learning model using Flask. We covered the process of setting up a Flask application, importing a pre-trained model, defining API endpoints, and deploying the model to a cloud platform. Now you can leverage Flask to create powerful machine learning APIs for production use.

I hope you found this tutorial helpful! If you have any questions or run into any issues, feel free to ask in the comments section below. Happy coding!

Published:

Related Articles