问题
I'm trying to insert a Pandas DataFrame into Mongodb using PyMongo.
df.head()
Because the index if the Pandas DataFrame is a DatetimeIndex, converting the DataFrame to a dict and inserting it to Mongodb:
db.testCollection.insert( df.T.to_dict() )
gives rise to an error:
InvalidDocument: documents must have only string keys, key was Timestamp('2016-04-07 09:30:00')
How can we convert the DatetimeIndex to something else that can be inserted into Mongodb, and later still be able to be converted back to a DatetimeIndex when reading from Mongodb?
回答1:
A solution would be that you turn index to str before attempting to store away in MongoDB, like this:
>> df.index = df.index.astype(str)
>> db.testCollection.insert(df.T.to_dict())
When reading the data out of db again later you can turn index to timestamp:
>> df.index = pd.to_datetime(df.index)
I hope this helps
来源：https://stackoverflow.com/questions/36902291/insert-pandas-timestamp-into-mongodb