EC2 Application ML Demo

• EC2 Application ML Demo
• Running Locally to generate the weights and encoders
  • Python Environment
  • Train the model and generate the files used in the application
  • Running application (locally)
• Run application on the open web (AWS)
  • Minor change on the app.py
  • EC2 – Instance
  • Security Group
  • Connecting with your virtual server machine
    • PuTTY
    • FileZilla (or WinSCP)
  • Installing dependencies on EC2
  • Running the Application
  • Open the web application

A simple machine learning application hosted on EC2 (AWS).

Running Locally to generate the weights and encoders

Python Environment

• Create the environment using virtualenv

$ python3 -m venv venv

• The next step is, of course, to activate the environment and install the dependencies:

$ source venv/bin/activate # unix
$ pip install -r requirements.txt

Train the model and generate the files used in the application

Now, you can run the script train.py without problem:

$ python train.py

This will generate three important files for the Dash Application in the root of the project folder:

EC2-APPLICATION-ML-DEMO/
├─ ...
├─ dummy_columns.pkl
├─ trained_model.pkl
├─ y_enc.pkl
├─ ...

Running application (locally)

After generating the trained model and auxiliary files, you need to run the app.py

$ python app.py

Type in your browser the generated link: localhost:8050 and you be able to see the application on the web.

Run application on the open web (AWS)

Minor change on the app.py

IMPORTANT: Go to the app.py and make the following changes (you just need to uncomment and comment):

# CODE HERE

if __name__ == "__main__":
    app.run_server(debug=False, host="0.0.0.0", port=8080)
    # app.run(debug=True)

This is required before upload to the EC2 Storage (you can change after inside the PuTTY using vi or vim).

EC2 – Instance

First, you need to create an instance on EC2 (Elastic Compute Cloud). I'm using the free tier for avoiding costs on my credit card (Amazon Linux 2023 AMI 2023.1.20230825.0 x86_64 HVM kernel-6.1). I selected on Instance Type: t2.micro (Family: t2 - 1vCPU - 1GB Memory).

On the Key pair name click on Create new key pair if you don't have one and select the following options:

Finally, click on Launch instance and wait until the instance starts.

Security Group

Open EC2 - Security Groups → Create security group → Inbound rules → Add rule:

After create the security group before, go to your instance of EC2 and manage the security groups to add the new security group policy:

On Associated security groups select your new security group and click on Add security group → Save

Connecting with your virtual server machine

I'm using two software programs here for establishing the connection with the EC2 Instance:

1. PuTTY - Connect to an SSH server
2. FileZilla - Download and upload files via FTP, FTPS and SFTP. If you use Windows you can use WinSCP

PuTTY

1. Session:

Host Name (or IP address): ec2-user@<Your Public IPv4 address>
Port: 22
Connection type: SSH (Telnet)

2. Connection → SSH → Credentials

Load your credentials (Key-Pair) created and storage outside your code

And click on Open to establish the connection with the EC2 Instance.

FileZilla (or WinSCP)

We'll use this to transfer files more easily with the EC2. Go to File → Site Manager and fill as shown below:

On Host, you put your Public IPv4 address used on the PuTTY. Click on Connect and if everything works well, you can see the files from the EC2

Now, let's go transfer our files to the EC2 instance. I created a folder called: demo-app/ using the FileZilla as you can see by the figure above.

You'll go to the file manager and transfer the following files and folders: src/, app.py, dummy_columns.pkl, requirements.txt, trained_model.py, and y_enc.pkl. You'll have something like this:

Installing dependencies on EC2

• Go back to the terminal PuTTY, and enter in your folder demo-app ($ cd demo-app)
• Run the commands to create the virtual environment and then activate the environment

$ python3 -m venv venv
$ source venv/bin/activate

• Install the dependencies using pip: $ pip install -r requeriments

• NOTE: if you obtain an error like IOError: [Errno 28] No space left on device. You need to change your pip cache by executing the following commands:

$ EXPORT TMPDIR=$HOME/tmp
$ mkdir -p $TMPDIR
$ pip install -r requeriments.txt

Running the Application

It's time for running the application, inside PuTTY run the command $ python app.py and will start the web application (dash/flask):

Open the web application

Go to the EC2 → Instances and select your instance. In the summary, you can see your Public IPv4 DNS copy the link and paste on the web browser, followed by the port: 8080.

<Public IPv4 DNS>:8080

Finally, you can see your application deployed on the EC2 - AWS (Image on the repository beginner)

---

"PEOPLE'S DREAMS WILL NEVER END" - OP

_{Created by BrenoAV}