SQL删除重复数据(根据多个字段)，pandas的nan存入数据库报错

delete from M_FACTOR_DATA_TEST a where (a.factor_id,a.data_date,a.stock_code) in (select factor_id,data_date,stock_code from M_FACTOR_DATA_TEST group by factor_id,data_date,stock_code having count(*) > 1)
and rowid not in (select min(rowid) from M_FACTOR_DATA_TEST group by factor_id,data_date,stock_code having count(*)>1)

当pandas的nan存入数据库报错是，想法是把Nan替换为None

df = df.where(df.notna(), None)

where回遍历df中的每个元素，判断notna（）时，用原本的元素填充（不变），遇到Nan时，用None替换