@chris441 Hey, sorry for not getting back to you. I didn’t solve it, but I exported the dataset as CSV and imported it into Python. I am on a tight schedule to produce some clustering for a researcher, so I didn’t want to risk not making it in time. If I manage to get it working in Python before I go on holiday, I will definitely attempt to backport it to Clojure (since I am much more comfortable using Clojure), but for something like this where I am not an expert (ML) using new Clojure libraries is a lot riskier than just using sklearn directly in Python.

This is the exact list I was looking for: https://github.com/cnuernber/dtype-next/blob/master/src/tech/v3/datatype/datetime/base.clj#L398-L416 I think this list should be present in your documentation. Magic keywords are quite hard to figure out for a beginner, especially when they’re not namespaced. It would make it a lot more beginner-friendly to to know what the legal typecasts are and which keywords are used to invoke them.

Again, amazing suite of libraries. I hope to get to dive deeper into them soon.

Great feedback - I completely agree

Also with decision. I would make same move. At least we now have the option to do similar in Clojure.