I\u2019ve been working on a side project and ended up compiling a dataset that may be useful beyond what I originally needed it for, so I\u2019m considering releasing it publicly.<\/p>\n
At a high level, the dataset contains:<\/p>\n
- \n
- structured records collected over a multi-year period<\/li>\n
- consistent timestamps and identifiers<\/li>\n
- minimal preprocessing (basic cleaning + deduplication only)<\/li>\n<\/ul>\n
It\u2019s not tied to a specific paper or product, more something that could support exploratory analysis, modeling, or benchmarking, depending on the use case.<\/p>\n
Before publishing, I wanted to sanity-check with this community:<\/p>\n
- \n
- what details do you usually look for to judge dataset quality?<\/li>\n
- is light preprocessing preferred, or raw + processed versions?<\/li>\n
- anything that would immediately make this more usable for research?<\/li>\n<\/ul>\n
Happy to share more specifics if there\u2019s interest, and open to feedback before release.<\/p>\n<\/div>\n
submitted by \/u\/crowpng <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
