Hi - I'm doing an integration of

daft

with my library

graphreduce

(https://wesmadrigal.github.io/GraphReduce/tutorial_swapping_compute/). It performs automated feature engineering, similar to

featuretools

but at larger scales due to integrations with spark and now, hopefully,

daft

. I'm not finding any clear way to perform window functions such as lag, lead, etc or even a similar pandas-esque shift (https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.shift.html). How can I perform windowing or even a simple shift in daft?

Hi Wes, Daft does not have windowing capabilities at the moment but it is something we are definitely looking to add. Do you have any suggestions for an API for this, or a specific operation that you would use?

Thanks for the response, Kevin. I don't have a strong suggestion for an API at the moment, but I'm a fan of spark's implementation. No rush, just was wondering. BTW, daft is fantastic.

Gotcha gotcha. We'll keep that in mind then. Glad to hear that there is interest in the community for window functions, it's something i've been wanting in Daft for a while as well, and maybe we'll bump its priority within the team!

Got

daft

integrated, regardless. Just won't be able to use window functions yet! Will update when those are in. https://github.com/wesmadrigal/GraphReduce/pull/17/files