This repository contains the analysis and experiments, which accompany the TüEyeQ data set and paper.

The dataset may be downloaded from Harvard Dataverse

Please refer to our nature article, if you are using the data set:

Kasneci, Enkelejda, et al. "TüEyeQ, a rich IQ test performance data set with eye movement, educational and socio-demographic information." Scientific Data 8.1 (2021): 1-14.

All experiments are provided in the Jupyter notebook analysis.ipynb. To reproduce the plots and tables shown in the paper, please download both the TüEyeQ data set and notebook and run all cells. For further information about the experiment's structure, please refer to the comments of the Jupyter notebook and the corresponding paper.

Pre-computed Distance Correlation scores can can be loaded from /pre-computed. Moreover, all plots from the paper can be found in /figures.

In future, we plan to extend this repository by a selection of benchmark predictive models. We also welcome any contributions from the community.