CP-Movetime-app-classifier

This repository contains the code for processing and classifying accelerometer data from binary files. The script processes raw accelerometer data, extracts features, and classifies the data using a trained Random Forest model. It also detects non-wear periods and labels changes between upright and non-upright body positions.

Setup

1. Clone the repository:

   git clone https://github.com/Ingemar28/CP-Movetime-app-classifier.git
   cd CP-Movetime-app-classifier

2. Install the required Python packages:

   pip install -r requirements.txt

Usage

1. Prepare your input data:

   • Place your binary .bin files in an input folder.
   • Ensure that the folder paths are correct when prompted by the script.

2. Run the script:

   python scripts/classify.py

3. The script will prompt you to enter the paths to your input and output folders:

   Please enter the path to your input folder containing .bin files: <path_to_input_folder>
   Please enter the path to your output folder: <path_to_output_folder>
   

4. The script will process each .bin file in the input folder, extract features, and classify the data. The results will be saved as CSV files in the output folder.

Detailed Steps in the Script

1. Read Binary Files:

   • The read_binary_file function reads binary files and converts them to a pandas DataFrame.

2. Smooth the Data:

   • The smoothing_data function smooths the accelerometer data using a rolling mean.

3. Segment Data and Extract Features:

   • The window_features function segments the data into windows and extracts features using the extract_acc_fet function.

4. Predict Using the Trained Model:

   • The script uses a pre-trained Random Forest model to predict the activity for each window of data.

5. Detect Non-Wear Periods:

   • The detect_non_wear_periods function identifies non-wear periods based on the standard deviation of the accelerometer data.

6. Label Changes Between Upright and Non-Upright Positions:

   • The script labels changes in body position (e.g., standing up, sitting down) based on the predicted activities.

7. Save the Results:

   • The results are saved to CSV files in the specified output folder.