Implement DataFrameGroupBy.count()

[sethmlarson]

No description provided.

[V1NAY8]

So, Ill implement the following:
Because pandas supports the same

>>> ed_df.groupby("dayOfWeek").count()
>>> ed_df.groupby("dayOfWeek").agg(["count"])

[sethmlarson]

I wonder if you can use doc_count within each bucket? Gotta check how pandas counts NaNs.

[V1NAY8]

Yeah, I thought of that.
While investigating it, I found that this is failing

# Flights index
ed_df.agg(['count'])
# Raises an exception

In the _map_pd_aggs_to_es_aggs

• we are mapping pd_agg 'count' to es_agg 'count'
• We have to actually map it as 'count' to 'value_count'
• This is a bug in aggregations

Does value_count work as expected for all type of fields?

[sethmlarson]

Seems like a potential bug to me :) If you wanna fix all in one go that'd be good. Im pretty sure value_count works for what we want.