将 function 应用于 Pandas 中的数据帧返回 UnboundLocalError

Question

My dataframe looks like this:

        Name    Overall Rating  Value in Millions
0   Neymar Jr   92  €105.5M
1   L. Messi    94  €95.5M
2   K. Mbappé   89  €93.5M
3   V. van Dijk     91  €90M
4   K. De Bruyne    91  €90M
...     ...     ...     ...
19692   I. Isa  63  €0
19693   I. Fetfatzidis  74  €0
19694   M. Mohsen   66  €0
19695   B. Jokič    72  €0
19696   B. Sigurðarson  73  €0

I am trying to apply a function to the 3rd column "Value in Millions" to convert values from string format to floats:

#A function to convert the values in the third row from strings to floats
    
def value_to_float(value_as_string): # eg.'€95.5M'
    
    value_as_string = value_as_string.strip('€')   
    
    if 'M' in value_as_string:                       #95.5M - string
        value_as_string = value_as_string.strip('M') #95.5 - string
        multiplier = float(value_as_string)          #95.5 - float
        value_as_float = multiplier * 1000000        #95000000.0 - float
        
    if 'K' in value_as_string:
        value_as_string = value_as_string.strip('K') 
        multiplier = float(value_as_string)
        value_as_float = multiplier * 1000           #Same as above, in case of K(Thousands)
        
    return value_as_float

The function works correctly when given an explicit parameter:

value_to_float('€95.5M')

95500000.0

However, when I try the following:

players["Value in Millions"].apply(value_to_float)

I get this error:

---------------------------------------------------------------------------
UnboundLocalError                         Traceback (most recent call last)
<ipython-input-80-3d3345f9405d> in <module>
----> 1 players["Value in Millions"].apply(value_to_float)

~/anaconda3/lib/python3.7/site-packages/pandas/core/series.py in apply(self, func, convert_dtype, args, **kwds)
   3846             else:
   3847                 values = self.astype(object).values
-> 3848                 mapped = lib.map_infer(values, f, convert=convert_dtype)
   3849 
   3850         if len(mapped) and isinstance(mapped[0], Series):

pandas/_libs/lib.pyx in pandas._libs.lib.map_infer()

<ipython-input-61-158745b17930> in value_to_float(value_as_string)
     15         value_as_float = multiplier * 1000           #Same as above, in case of K(Thousands)
     16 
---> 17     return value_as_float

UnboundLocalError: local variable 'value_as_float' referenced before assignment

I tried several other methods(series.map(), oldschool looping), but I always get the same error, so I'm inclined to think there's a gap in the logic somewhere.

Answer 1

Your issue stems from the fact that your function assumes that all rows have an "M" or "K" in their values for "Value in millions". In your own dataframe example above we can clearly see examples where the "Value in millions" is "€0". Following the function logic, the value_as_float variable never gets set so it throws the error you outlined.

Adjusting your function to set the value_as_float to 0 by default fixes this issue.

def value_to_float(value_as_string): # eg.'€95.5M'
    
    value_as_string = value_as_string.strip('€')
    value_as_float = 0
    
    if 'M' in value_as_string:                       #95.5M - string
        value_as_string = value_as_string.strip('M') #95.5 - string
        multiplier = float(value_as_string)          #95.5 - float
        value_as_float = multiplier * 1000000        #95000000.0 - float
        
    if 'K' in value_as_string:
        value_as_string = value_as_string.strip('K') 
        multiplier = float(value_as_string)
        value_as_float = multiplier * 1000           #Same as above, in case of K(Thousands)
        
    return value_as_float