Project description

This project is a Flask-based web application designed to predict the class a user falls into (1-5) based on their mobile device usage. The application exposes a single endpoint /predict which accepts POST requests containing user data in JSON format. The user data includes various features related to mobile device usage, such as app usage time, screen-on time, battery drain, number of apps installed, data usage, age, gender, and user behavior class.

The prediction is made using a pre-trained machine learning model, which is invoked by the predict function. The result of the prediction is returned as a JSON response.

The application is designed to be run using the waitress WSGI server for production deployment.

Endpoint

• POST /predict
  • Request Body: JSON object containing user data with the following fields:
    • device_model: The model of the user's mobile device.
    • operating_system: The operating system of the user's mobile device.
    • app_usage_time(hours/day): The average app usage time per day in hours.
    • screen_on_time(hours/day): The average screen-on time per day in hours.
    • battery_drain(mAh/day): The average battery drain per day in mAh.
    • number_of_apps_installed: The number of apps installed on the user's device.
    • data_usage(mb/day): The average data usage per day in MB.
    • age: The age of the user.
    • gender: The gender of the user.
    • user_behavior_class: The user behavior class.
  • Response: JSON object containing the predicted class.

Setting Up

Clone the repository

git clone https://github.com/Jaykold/user-behavior-classification.git

Navigate to the project directory

cd user-behavior-classification

Create a virtual environment

python -m venv myenv

After creating the virtual environment, you need to activate it:

• On Windows:

myenv\Scripts\activate

• On macOS and Linux:

source myenv/bin/activate

You can install the project dependencies by running this command:

pip install -r requirements.txt

Running the Application

To run the application, execute the following command:

waitress-serve --listen=0.0.0.0:9696 main:app

Building & Running the Docker container

1. Build the docker image using this code

docker build -t user-classifer .

2. Run the container

docker run -it --rm -p 9696:9696 user-classifier

3. Test the running container with this command:

python test_url.py

4. Push to DockerHub (Optional)

docker tag user-classifier <dockerhubName/user-classifier:latest>