使用带有列的 DataFrame 作为 str.format() 的命名参数

Question

我有一个像这样的 DataFrame：

import pandas as pd
df = pd.DataFrame({'author':["Melville","Hemingway","Faulkner"],
                   'title':["Moby Dick","The Sun Also Rises","The Sound and the Fury"],
                   'subject':["whaling","bullfighting","a messed-up family"]
                   })

我知道我可以做到：

# produces desired output                   
("Some guy " + df['author'] + " wrote a book called " + 
   df['title'] + " that uses " + df['subject'] + 
   " as a metaphor for the human condition.")

但是否可以使用 str.format() 将其写得更清楚，类似于：

# returns KeyError:'author'
["Some guy {author} wrote a book called {title} that uses "
   "{subject} as a metaphor for the human condition.".format(x) 
      for x in df.itertuples(index=False)]

Answer 1

>>> ["Some guy {author} wrote a book called {title} that uses "
   "{subject} as a metaphor for the human condition.".format(**x._asdict())
      for x in df.itertuples(index=False)]

['Some guy Melville wrote a book called Moby Dick that uses whaling as a metaphor for the human condition.', 'Some guy Hemingway wrote a book called The Sun Also Rises that uses bullfighting as a metaphor for the human condition.', 'Some guy Faulkner wrote a book called The Sound and the Fury that uses a messed-up family as a metaphor for the human condition.']

请注意 _asdict() 并不是 public api 的一部分，因此依赖它可能会在以后对 pandas.[=13 的更新中中断=]

您可以这样做：

>>> ["Some guy {} wrote a book called {} that uses "
   "{} as a metaphor for the human condition.".format(*x)
      for x in df.values]

Answer 2

您也可以这样使用 DataFrame.iterrows()：

["The book {title} by {author} uses "
   "{subject} as a metaphor for the human condition.".format(**x) 
     for i, x in df.iterrows()]

如果你想，哪个很好：

使用命名参数，因此使用顺序不必与列的顺序匹配（如上）
不使用像_asdict()

Timing：最快的似乎是M. Klugerford的第二个解决方案，即使我们注意到关于缓存的警告并采取最慢的运行。

# example
%%timeit
 ("Some guy " + df['author'] + " wrote a book called " + 
   df['title'] + " that uses " + df['subject'] + 
   " as a metaphor for the human condition.")
# 1000 loops, best of 3: 883 µs per loop

%%timeit
    ["Some guy {author} wrote a book called {title} that uses "
       "{subject} as a metaphor for the human condition.".format(**x._asdict())
          for x in df.itertuples(index=False)]
#1000 loops, best of 3: 962 µs per loop

%%timeit
    ["Some guy {} wrote a book called {} that uses "
     "{} as a metaphor for the human condition.".format(*x)
          for x in df.values]   
#The slowest run took 5.90 times longer than the fastest. This could mean that an intermediate result is being cached.
#10000 loops, best of 3: 18.9 µs per loop

%%timeit
    from collections import OrderedDict
    ["The book {title} by {author} uses "
       "{subject} as a metaphor for the human condition.".format(**x) 
         for x in [OrderedDict(row) for i, row in df.iterrows()]]
#1000 loops, best of 3: 308 µs per loop            

%%timeit 
    ["The book {title} by {author} uses "
       "{subject} as a metaphor for the human condition.".format(**x) 
         for i, x in df.iterrows()]
#1000 loops, best of 3: 413 µs per loop

为什么倒数第二个比最后一个快我搞不懂。

使用带有列的 DataFrame 作为 str.format() 的命名参数

using a DataFrame with columns as named arguments to str.format()

python

string-formatting

pandas