This is the analysis submodule repository of the dl_har_public repository.
If you want to contribute to this repository make sure to fork and clone the main repository dl_har_public repository with all its submodules. To do so please run:
git clone --recurse-submodules -j8 git@github.com:STRCSussex-UbiCompSiegen/dl_har_public.git
If you want to have your modification be merged into the repository, please issue a pull request. If you don't know how to do so, please check out this guide.
In the following each of the main components will be briefly summarised. Output of the analysis
Contains all relevant methods to analyse (previously saved) training and testing predictions. Depending on the type of validation method employed during training results are also printed out aggregated subject-wise.
The analysis.py script can also be run on its own making it possible to rerun the analysis of a previously run experiment. To do so run:
python analysis.py -d [timestamp of the experiment]
Note that the timestamp of the experiment is to be written in the format YYYYMMDD/hhmms (which is equivalent to the way the log directory is structured)