patrickhulce
Add IQR and MAD to execution time summary
Great work!
Leaving this as a note for myself or for anyone who can contribute.
Seeing the IQR and MAD could give us a better idea of intra-library variation. It would be great to see how analytics library execution times spread out in this dataset.
I suspect that the average creeps up with certain libraries due to the population that uses them. For instance, the developers that integrate them may be more skilled, and hence use more feature of the library.
Good point @phaseOne! I'll consider this in a future update. The major blocker right now would be querying. Right now everything can be done in one go with a GROUP BY basically, but the bigquery QUANTILES function can't be used in combination with this AFAIK. If you know any ways around this or someone that has tips, very open to recommendations :)
