An Easy Guide on Machine Learning Model Deployment Using Heroku (Flask)

 

 

 

Heroku offers many ways of deploying your applications to Heroku.In this post, we will use GitHub Integration.You add your project to a GitHub repository. Then you can deploy changes directly to Heroku by linking Heroku to GitHub. We will see the steps in details.

So this post focuses in the deployment phase, I suppose that you create your model and set up your web app.

Adding Files

After creating your model and setting up your web app. Your Folder should be like this :

•  static: for css files

• template: for html files

• app.py :the main code of the app

• CAR_DETAILS.csv:  data

• model.pkl: model saved using pickle

• model.py: for the model creation and training

We need to add some files to the folder :

Step1: Add a Procfile file

go to your project and add a Procfile file.It should be without extension like "Procfile.txt" , just "Procfile"
write inside the file the following code

Procfile code

Step2: Add a requirements.txt

It is a text file that contains the python packages required to execute the application.It's recommendable to  mention the versions of all your libraries.

 

Step3:Committing Code in a GitHub repository:

 

 

Step4: Deploy the app on Heroku 

1. click on Create new app

   

   
    

2. Enter Your app name and Click Create

   
   
   

   
   

3. click on Connect to GitHub

   

   
    

4.  choose the repository where you uploaded the project  code  in the search, Click connect

   

   
    

5.   Click deploy Branch

   

   
   

6.  your app was successfully deployed

   

   
    

7. Testing your app

   click view app
   
   The app is published at URL:https://heruko-deployment-app.herokuapp.com/  

   
   

   You can refer to my GitHub repository for this project.