运行以下代码时，我收到此错误“分配前引用的局部变量 'col'”：

Question

def log_hist(data):
    data = pd.DataFrame([])
    for col in data:
        data_log = np.log(data[col])
    return data[col].hist()




---------------------------------------------------------------------------
UnboundLocalError                         Traceback (most recent call last)
<ipython-input-127-dc391683ef40> in <module>()
----> 1 log_hist(df2)

<ipython-input-126-60d8684e3abe> in log_hist(data)
      3     for col in data:
      4         data_log = np.log(data[col])
----> 5     return data[col].hist()
      6 
      7 

UnboundLocalError: local variable 'col' referenced before assignment

Answer 1

Since your dataframe is empty, there is nothing to iterate through. The for loop does not instantiate the col variable. If your iterable (the dataframe) was not empty, then the variable would be available after the loop.

Example of idea:

>>> for x in range(10):
...  pass
... 
>>> x
9
>>> for y in []:
...  pass
... 
>>> y
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'y' is not defined
>>> 

Answer 2

The variable col is only valid within the for loop. If you want to access the value outside of the loop (eg to capture the last value that is processed by the loop), declare a variable outside of the loop:

def log_hist(data):
    last_col = None
    for col in data:
        np.log(data[col])
        last_col = col
    return data[last_col].hist()  # will raise an exception if the dataframe is empty!

Note that I removed the line at the start of your function which zeroes out the data parameter -- I assume this was part of some debugging attempt, but it obviously completely breaks the function to have that in there. :)

I also removed the data_log variable since it's not used anywhere; there's no point assigning a variable that you never use. (I'm not familiar with np.log , I assume it has a side-effect that is desirable -- otherwise the loop is mostly pointless to begin with and should be replaced with code that simply selects the last column without looping through the entire object.)