How to rename a Pandas DataFrame column if the name is an empty string?

Question

I've imported an HTML table from Basketball Reference using pandas, but I'm running into an annoyance trying to rename a couple of columns that have empty strings for their name.

Here's the code to pull the table:

tables = pd.read_html('http://www.basketball-reference.com/leagues/NBA_2016_games.html')
games = tables[0]

The columns look like this:

Out[138]: 

Index([u'Date', u'Start (ET)', u'Visitor/Neutral', u'PTS', u'Home/Neutral',
       u'PTS.1', u' ', u' .1', u'Notes'],
       dtype='object')

Renaming everything except for the u' ' and u' .1' columns is no issue, but I cannot find the right way to rename the empty ones using a label approach.

I tried this by default (limited to renaming only a few columns here):

column_names = {'Date': 'date', ' ': 'box', ' .1': 'overtime'}
games.rename(columns = column_names)

but this leaves the ' ' and ' .1' columns unchanged.

This method works:

column_names = {games.columns[6]: 'box', games.columns[7]: 'overtime'}

But is there any way to change these names without explicitly referencing the position?

Answer 1

也许这可能是一个快速修复 - 明确设置列名称。

df.columns = [u'Date', u'Start (ET)', u'Visitor/Neutral', u'PTS', u'Home/Neutral', u'PTS.1', u'Rename1', u'Rename2', u'Notes']

Answer 2

For me works add str.strip for remove trailing whitespaces, also is necessary change dict values (remove whitespaces):

column_names = {'Date': 'date', '': 'box', '.1': 'overtime'}
games.columns = games.columns.str.strip()
games = games.rename(columns = column_names)
print (games.columns)
Index(['date', 'Start (ET)', 'Visitor/Neutral', 'PTS', 'Home/Neutral', 'PTS.1',
       'box', 'overtime', 'Notes'],
      dtype='object')

Another solution is export column names to list and there is \\xa ( NO-BREAK SPACE ):

print (games.columns.tolist())
['Date', 'Start (ET)', 'Visitor/Neutral', 'PTS', 'Home/Neutral', 
 'PTS.1', '\xa0', '\xa0.1', 'Notes']


column_names = {'Date': 'date', '\xa0': 'box', '\xa0.1': 'overtime'}
games = games.rename(columns = column_names)
print (games.columns)
Index(['date', 'Start (ET)', 'Visitor/Neutral', 'PTS', 'Home/Neutral', 'PTS.1',
       'box', 'overtime', 'Notes'],
      dtype='object')