The major contribution of the article is the use of autoencoders to make high-dimensional movement data comprehensible for designers. We aim at bringing the machine-learning team and interaction design team together and build a common ground for discussion. For this reason, we want to share our code and data with the readers and hope to support future design processes.
The following figures contain Python notebooks that can be used to experiment with the provided data and autoencoder models. In case of any runtime issues with the current Python environment, we additionally offer access to our repository containing all code, data and models mentioned in the article. The repository also contains copies of the Python notebooks linked to the figures.