np.char.split returns an array of lists, not a 2d array

[Msegade]

The returned array of the function np.char.split gives an array of list, which has to be further processed to produce a normal 2d numpy array

Example

>>> a = np.array([['0.1 0.2 0.3'], ['0.3 0.4 0.5'], ['0.5 0.6 0.7']])
>>> np.char.split(a)
array([[list(['0.1', '0.2', '0.3'])],
       [list(['0.3', '0.4', '0.5'])],
       [list(['0.5', '0.6', '0.7'])]], dtype=object)

Solution

To ease the processing of the arrays, it will be better to have the function return

array([['0.1', '0.2', '0.3'],
       ['0.3', '0.4', '0.5'],
       ['0.5', '0.6', '0.7']], dtype='<U3')

[eric-wieser]

What would you have the following produce?

>>> a = np.array([['0.1 0.2'], ['0.3 0.4 0.5 0.6']])
>>> np.char.split(a)

[Msegade]

Ok, I see the issue. Maybe a function argument that generates the 2d array if the resulting lists have the same length, and otherwise raises an exception?

np.char.split(a, homogeneous_array=True)

[amjangde]

First convert the array (i.e. array of lists returned by the split() method) to list. So now we have list of lists. And now use 'np.array()' method to convert the list (i.e. list of lists) to a 2D array.

Please refer this converty numpy array of arrays to 2d array

example:

date_cols = ['2012-12-12', '2011-11-11', '2010-10-10']

date_cols = np.char.split(date_cols,'-')

//here we will get array of lists
//date_cols will look like -> [list(['2012', '12', '12']), list(['2011', '11', '11']), list(['2010', '10', '10'])]

date_cols = date_cols.tolist()
date_cols = np.array(date_cols)

//date_cols will look like this -> [['2012' '12' '12'] ['2011' '11' '11'] ['2010' '10' '10'] which is a 2D array

[eric-wieser]

A way to solve this is to use:

def array_of_lists_to_array(arr):
    return np.apply_along_axis(lambda a: np.array(a[0]), -1, arr[..., None])

Obviously this spelling is less than ideal, but it does correctly throw an exception if something is amiss.

I'm going to close this issue, I doubt we are in favor of adding special-cases to char.split. I've opened #14478 to track adding a generic helper to solve this kind of problem.