Force pandas to interpret (1,2) in column as strin

问题:

I have this weird behaviour in a pandas Dataframe. I am using .apply(single_seats_comma) on a column with the following example content: (1,2). However, it seems to return it as range(1,3) instead of a string (1,2). Other rows have more than 2 entries as well, e.g. (30,31,32). I have a function which splits on , and converts each value in brackets into a new row however with (x,x) it breaks.

def single_seats_comma(row):
    strlist = str(row).split(',')
    strlist = filter(None, strlist) 
    intlist = []
    for el in strlist:
        intlist.append(int(el))
    return intlist

Example for 'apply':

tickets['seats'][:1].apply(single_seats_comma)

The Error output of the def is

ValueError: invalid literal for int() with base 10: 'range(1'

Trying to find a solution, I found this:

str(tickets['seats'][:1])
>>'0    (1, 2)\nName: seats, dtype: object'

tickets['seats'][:1].values
>> '[range(1, 3)]'

It works on a column if the values are just 1,2.

Any help help is much appreciated!

回答1:

Perhaps it would be easier to simply iterate over the elements of the row instead of converting to string then splitting. This is simple enough to use a lambda.

tickets['seats'][:1].apply(lambda row: [int(e) for e in row])

回答2:

I cannot reproduce the rangestring. But this function should work for both cases:

def single_seats_comma(row):
    if type(row) is tuple:
        return list(row)
    elif type(row) is range:
        res = [row.start]
        end = row.stop - 1
        if end - row.start > 1:
            res.append(end)
    return res

Example:

>>> tickets = pd.DataFrame({'seats': [(100, 1022), range(3, 4), range(2, 10)]})
>>> tickets['seats'].apply(single_seats_comma)
0    [100, 1022]
1            [3]
2         [2, 9]
Name: seats, dtype: object

回答3:

Thanks to all contributors to get me closer to a solution. The solution is actually quite simple.

The challenge was that pandas interpreted (1,2) as range and not as string However, the target was to create a list of all values, originally by splitting a string on ','. Not needed!

list(range(1,2)) does the job already. Here is the example and solution:

list(range(11, 17))
>> [11, 12, 13, 14, 15, 16]

tickets['seats'][0]
>> range(1, 3)

list(alltickets['seats'][0])
>> [1, 2]

So solution(s):

def single_seats_comma(row):
    strlist = list(row)
    return strlist

tickets['seats'].apply(single_seats_comma)

or

tickets['seats'].apply(lambda row: list(row))