For example I have this pandas simple data frame:
>> print(file)
Name
0 ['junior','senior']
1 freshgrad
when I tried using :
>> len(file[0])
2
but for the second one
>> len(file[1])
9
But I want the second one to calculate as 1 how to differentiate between these two?
.join() but it still calculate as 9list.count but I got an errorThe best way to do what you want is to check the data type of the item in question. You can use:
if isinstance(item, list):
...
And:
if isinstance(item, str):
...
In the case of a string, you can then use 1 for the length if you wish. Note that it's better to use isinstance(item, dtype) than type(item) == dtype because it will automatically work on subclassed types.
freshgrad is a string
so len(file[1]) means that you are calculate the length of this string.It is 9.
if file[1] is a list containing freshgrad like['freshgrad'],len(file[1]) will be 1.
You could use np.size:
In [301]: file = pd.Series([['junior','senior'], 'freshgrad'])
In [302]: file.apply(np.size)
Out[302]:
0 2
1 1
dtype: int64
In [327]: np.size(file[0])
Out[327]: 2
In [328]: np.size(file[1])
Out[328]: 1
But to some extent this might just be delaying your agony. When the objects in
a Series (or any kind sequence) have different types, the code tends require type-checking or try..excepts to handle the various types differently. (In fact, this is what np.size is doing. Under the hood np.size is using try..except to handle the exceptional case.)
Life is usually simpler (and therefore better) when all the objects in a sequence have the same type. So it is preferable to build a Series whose elements are all lists:
In [301]: file = pd.Series([['junior','senior'], ['freshgrad']])
You could define your own length function, which uses the type to determine how to calculate the length:
def mylen(data):
return len(data) if isinstance(data,list) else 1
1 ['freshgrad']you would getlen(file[1])=1freshgradis currently a string, and you want it to be an array.