Aggregation over multiple fields

[TeweiLuo]

For example, is something like

res = sp.partial_sql("SELECT ticker, trade_day min(price), max(price) FROM table GROUP BY ticker, trade_day", df)

supported?

Also we see in the README

df = df.repartition(3, hash_by="ticker")

which suggests that we need to manually shuffle the data before an aggregation. Is this strictly required?

[wangrunji0408]

Yes. You can write anything in SQL, but it is only processed within each partition. So a manual repartition is required before aggregation.