The project's goal is to compare performance of various machine learning techniques in recognizing activity performed by a human based on IMU readings on their phone.
Project was developed with python 3.11.0
To get started, create a virtual environment,
python -m venv venv
and activate it with the script matching you operating system.
Install required packages
pip install -r requirements.txt
You can interact with the data and models using jupyter notebooks
- Data visualization
- Data preparation
- Models
- Logistic Regression, KNeighbors, SVC, DecisionTree, RandomForest
- Long short-term memory (LSTM)
Additionaly, time_per_activity.py helps with summarizing total time for each activity type. grid.py enables the search of hiperparameter space, and hiperparam_summary.py gathers it's results.