Skip to content

Latest commit

 

History

History
84 lines (57 loc) · 2.82 KB

README.md

File metadata and controls

84 lines (57 loc) · 2.82 KB
Raw

Body Performance Predictor Docker Application

This is a dockerized application that uses machine learning to predict body performance based on physical and lifestyle data. It was made by Tommaso Martinelli and Paolo Speziali for the "Virtual Networks and Cloud Computing" exam at UniPG.

Machine Learning Model

The machine learning model used in this application was trained in a Jupyter Notebook. The code for this notebook can be found in the Body-Performance-Predictor repository.

Getting Started

To run this application, you must have Docker and Docker Compose installed on your machine. If you do not have these, please follow the installation instructions for your operating system:

To install the application navigate into the application root directory and execute the script:

chmod u+x init.sh
./init.sh

If you prefer to use Kubernetes instead of Docker Compose, execute this:

chmod u+x init-k.sh
./init-k.sh

and then, on two different terminals, write the following:

kubectl port-forward service/webserver 80:8080

and

kubectl port-forward service/pgadmin 8081:8081

Usage

To predict body performance, fill out the form on the home page with your personal data and click "Predict". The application will use machine learning to calculate your body performance score. You may also add more data, the model will automatically train again.

You can also access the PgAdmin web interface by visiting http://localhost:8081 in your web browser. The default login credentials are:

Docker Containers Configuration

This application consists of the following Docker containers:

  • webserver: the web application
  • predictor: the machine learning prediction service
  • training: the machine learning training service
  • inserter: the service responsible for inserting new records into the database
  • initializer: the service responsible for initializing the database with the training data and query the training service for the first time
  • db: the PostgreSQL database
  • pgadmin: the PgAdmin web interface

Ports

The application uses the following ports:

  • 80: the web application
  • 5000: the prediction service
  • 3002: the training service
  • 3001: the inserter service
  • 5432: the database
  • 8081: the PgAdmin web interface

Volumes

The application uses the following volumes:

  • postgres_data: the PostgreSQL database data directory
  • model_data: the machine learning model data and scaler model data directory

Networks

The application uses a bridge network named app_network.