获取 pandas dataframe 中的所有列，当不同列中存在不同时区时，该列是日期列

Question

(Note, this SO question does not take into account the different time-zones ie it does not catch a date when a time-zone is present).

I have a dataframe where some column dtypes are datetime64[ns] and some are datetime64[ns,UTC] .

Note, they have all been converted using

df["some_time_col"] = pd.to_datetime(df["some_time_col"]) but since df is a merge of data across different databases, the time-zones are different.

Is there a way to get all columns that are a date-column ie something like this

dt_cols = df.select_dtypes(include=[pd.datetime]).columns

instead of having to specify all time-zones like

dt_cols = df.select_dtypes(include=["datetim64","datetime64[UTC]","datetime64[UTC+1]"...).columns

Using either of

dt_cols = df.select_dtypes(include=[np.datetime64]).columns
dt_cols = df.select_dtypes(include=["datetime64"]).columns

does not catch datetime64[ns,UTC] but only datetime64[ns]

Answer 1

Use:

df.select_dtypes('datetimetz')

From select_dtypes docs:

To select Pandas datetimetz dtypes, use 'datetimetz' (new in 0.20.0) or 'datetime64[ns, tz]'

For selecting both datetime without tz and with tz, do:

df.select_dtypes(['datetimetz', 'np.datetime64'])