這是 Pandas 等同於排序分區上的 SQL 密集排名嗎？

Question

我有以下 SQL 片段，我想在 Python 中復制：

SELECT
   DENSE_RANK() OVER (PARTITION BY a0.[field1], a0.[field2] ORDER BY a0.[field3] DESC)  AS [new_field]
   ...
FROM <table>

這是我的嘗試，這等價嗎？

df["new_field"] = df.groupby(["field1", "field2"])["field3"].rank(
    method="dense", ascending=False
)

Answer 1

是的，只要field3是數字就等價。 否則你需要一些操作，比如

df['new_field'] = df.field3.rank(method='dense',ascending = False)
df['new_field'] = df.groupby(['field1','field2'])['new_field'].rank(method='dense').astype(int)