Skip to content

A machine learning-based application for user action recognition using smartphone sensors, showcasing data analysis and predictive modeling in context-sensitive systems.

eismont21/context-aware-ml-app

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

30 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Context-Sensitive Application

Overview

Developed as part of the Context Sensitive Systems lecture series at KIT, this application uses machine learning to demonstrate the practical applications of context recognition in mobile devices. It effectively utilizes smartphone sensor data to predict user actions.

Components

Record App

  • Collects accelerometer data from smartphones for three activities:
    • Reading: The phone is held stationary, typically in a position for reading.
    • Chatting: Similar orientation to reading, but with intermittent typing.
    • Calling: The phone is held close to the ear, simulating a call.
  • Data is uploaded to the edge-ml cloud, demonstrating real-world data collection in context-sensitive systems.

Python Data Analysis with ML Pipeline (Notebooks)

  • The project includes Jupyter Notebooks covering the full spectrum of a ML pipeline:
    • data cleaning → feature extraction → feature selection → model selection → hyperparameter optimization → deployment

sayMyActionApp

  • A web application that uses the developed model to identify and predict the user's current action based on sensor data.
  • Test the app here.

Getting Started

  • Setting Up the Environment:
    • Use the provided env.yml file to create a conda environment with all required dependencies.
  • Running the Applications:
    • For the Record App, follow the instructions in RecordApp/README.md.
    • The sayMyActionApp can be tested directly from the provided CodeSandbox link.
  • Accessing the Notebooks:
    • The Jupyter Notebooks are located in the notebooks directory. They can be run to replicate the data analysis and model training processes.

Note: The application has been tested only on iOS devices.

About

A machine learning-based application for user action recognition using smartphone sensors, showcasing data analysis and predictive modeling in context-sensitive systems.

Topics

Resources

Stars

Watchers

Forks

Languages

  • Jupyter Notebook 99.8%
  • Other 0.2%