Any suggestions for new datasets are welcome. Please raise an issue or PR. Any datasets used should be in the public domain, or provided with attributions.