The Arrow C++ library has tdigest and hash_tdigest kernels that return arrays/lists. We call these from bindings with q = 0.5 to find the approximate median, but the output is difficult to work with. It would be nice to create kernels wrapping these that would produce scalar output per array/group:

appx_median --> tdigest(q = 0.5)

hash_appx_median --> hash_tdigest(q = 0.5)

Reporter: Ian Cook / @ianmcook
Assignee: David Li / @lidavidm

Related issues:

• [R] Binding for median() and quantile() aggregation functions (blocks)

PRs and other links:

• GitHub Pull Request #11204

_{Note: This issue was originally created as ARROW-14052. Please see the migration documentation for further details.}