Python Pandas日期时间和多索引问题

Question

I have a Python script. After running various commands to import, transpose and process data from a CSV file, I end up with a dataframe that looks like this:

        PV          PV
Date    30/11/2016  01/12/2016 
00:30   4           4
01:00   5           1
01:30   6           7
etc

What I want now is to remove the column for 30/11/2016, leaving only the data for 01/12/2016. This is the code I have:

# create MultiIndex.from_arrays from first row of DataFrame first, then remove first row 
# by df.iloc
df.columns = pd.MultiIndex.from_arrays([df.columns, pd.to_datetime(df.iloc[0])])
df = df.iloc[1:]

# get today's date minus 60 mins. the minus 60 mins will account for the fact that the
# very last half hourly data slot is produced at the beginning of the next day
date = dt.datetime.today() - dt.timedelta(minutes=60)

# convert to correct format:
date = date.strftime("%d-%m-%Y")

# Use indexslice to remove unwanted date columns i.e. none that are not for today's 
# date
idx = pd.IndexSlice
df = df.loc[:,idx[:,[date]]]

# drop the second level of the multiindex, which is the level containing the date, which 
# is no longer required
df.columns = df.columns.droplevel(1)

This was working fine for the whole of November until today, the 1st December, when it started throwing up errors. What I've traced it to is the first section of code ie:

# create MultiIndex.from_arrays from first row of DataFrame first, then remove first row 
# by df.iloc
df.columns = pd.MultiIndex.from_arrays([df.columns, pd.to_datetime(df.iloc[0])])

The output of which is:

        PV         
Date    2016-11-30  2016-01-12
Date    30/11/2016  01/12/2016 
00:30   4           4
01:00   5           1
01:30   6           7
etc

The problem is in the first set of dates shown above, the first of which is 2016-11-30, therefore YMD, the second is 2016-01-12, therefore YDM. Why are the date formats different? How would I keep them both as YMD?

Answer 1

这有效：

df.columns = pd.MultiIndex.from_arrays([df.columns, pd.to_datetime(df.iloc[0], format='%d/%m/%Y')])