P99 custom aggregation

can someone show me an example of how to do p99 as an aggregation for each group?
I have an idea where I return the SeriesGroupBy object for each “chunk” (identity function) and in the “agg” function, I just try to concatenate all the chunks and perform p99

Hi @RAY2L, welcome to Dask community!

Your post is a little short, but I think that you are maybe looking for GroupBy/apply calls. This will automatically shuffle all the data belonging to one group into one single chunk on which you can perform Pandas operations.