This is the source code for "Model Evaluation Approaches for Human Activity Recognition from Time-Series Data" that was presented at AIME2021, June 17, 2021. https://doi.org/10.1007/978-3-030-77211-6_23
The input to most of the files are X, y, sub(ject) numpy arrays that have been stored to a mapped filesystem. These arrays are generated from the raw data by the code located in https://github.com/imics-lab/load_data_time_series
MobiAct_data_explore.ipynb was used to generate the values for the resampling and component versus total acceleration experiments.
model_eval_HAR_stratification.ipynb was used for the stratification experiments which (incorrectly) mixes subject data into the train and test sets.
model_eval_HAR_defined_subject.ipynb was used for the experimentation using subjects allocated to only one of train/test/validate.
subject_split_generator.ipynb was used to generate dictionaries containing subject splits that could be input into the load_data_time_series methods.
Plots_for_model_eval.ipynb was used to generate the plots that were not included in other files directly.
If you find this work useful please cite using this BibTeX
@inproceedings{hinkle2021model,
title={Model Evaluation Approaches for Human Activity Recognition from Time-Series Data},
author={Hinkle, Lee B and Metsis, Vangelis},
booktitle={International Conference on Artificial Intelligence in Medicine},
pages={209--215},
year={2021},
organization={Springer}
}