Appendix L — chapter4.pipeline.02_splitting-evaluation
chapter4.pipeline.02_splitting-evaluation
Splitting approach evaluation script
This script trains 100 models for each data source (smartphone, smartwatch) and splitting approach. For the training process, a 80/20 train/test split is employed with a batch size of 20 windows during 50 epochs.
Example:
$ python 02_splitting-evaluation.py
--ts_data_path <PATH_OF_TURNING_SITTING_DATA>
--tts_data_path <PATH_OF_TURN_TO_SIT_DATA>
--reports_output_path <PATH_TO_STORE_REPORTS>