This is a cleaned version of the experiment I developped during my Master's Thesis to detect falls based on wereable sensors. It resulted in two publications:

• Zurbuchen, N., Wilde, A., & Bruegger, P. (2021). “A Machine Learning Multi-Class Approach for Fall Detection Systems Based on Wearable Sensors with a Study on Sampling Rates Selection”. In: Sensors 21.3. ISSN: 1424-8220. DOI: 10.3390/s21030938.
• Zurbuchen, N., Bruegger, P., & Wilde, A. (2020). “A Comparison of Machine Learning Algorithms for Fall Detection using Wearable Sensors”. In: 2020 International Conference on Artificial Intelligence in Information and Communication (ICAIIC), pp. 427–431. DOI: 10.1109/ICAIIC48513.2020.9065205.

This experiment only works with the publicly available SisFall dataset which is a a fall and movement dataset based on wearable sensors.

The experiment is in the file main_experiment.py which requires the following input parameters:

• dataset_folder : The path of the folder containing the SisFall data set.
• output_folder : The path of the folder where all the results will be saved.

The following list defines the optional parameters which all have default values:

• -se, --sensors : The list of sensors axes as numbers from 0 to 8 included.
• -is, --ignored_subjects : The list of ignored subjects as subjects names from SA01 to SA23 and SE01 to SE15.
• -du, --duration : The duration of the sample in [ms] as a number between 1000 and 12000 included.
• -fr, --frequencies : The list of frequencies of the sampling [Hz] as numbers from 1 to 200 included and divisor of 200.
• -pr, --pre_time : The duration after the impact in [ms] (must be between 100 and 5000, only available with multi-class).
• -po, --post_time : The duration before the impact in [ms] (must be between 100 and 5000, only available with multi-class).
• -cl, --classification : The classification type (either binary or multi-class).
• -mo, --models : The list of machine learning algorithms to use (either knn, svm, dt, rg or gb).
• -kf, --k_fold : The number of folds to use (must be between 2 and 10).