[FIXED] Convert numpy array to pandas dataframe with labels from list

Issue

please advise how to perform the following permutations:

array = [1, 3, 2]  # numpy.ndarray

l1 = ['foo_qwe1_ert1', 'bar_qwe2_ert2', 'baz_qwe3_ert3']  # list

I need to get the following pandas dataframe:

Column1	Column2	Column3
foo	qwe1	ert1
baz	qwe3	ert3
bar	qwe2	ert2

the problem is the list contains text labels from 0 to 30 (format: XXX_YYY_ZZZ) and numpy.array has shape (3536,) and contains numbers from 0 to 30. I need to assign label for each number in array and save it as pandas dataframe

Solution

First use DataFrame constructor with split:

df = pd.DataFrame([x.split('_') for x in l1], columns=['Column1', 'Column2', 'Column3'])
print (df)
  Column1 Column2 Column3
0     foo    qwe1    ert1
1     bar    qwe2    ert2
2     baz    qwe3    ert3

And then change order by array by extract last integer from last column:

df.index = df['Column3'].str.extract('(\d+)$', expand=False).astype(int)
df = df.loc[array].reset_index(drop=True)
print (df)
  Column1 Column2 Column3
0     foo    qwe1    ert1
1     baz    qwe3    ert3
2     bar    qwe2    ert2

EDIT:

array = np.array([1, 3, 2])
l1 = ['foo_qwe1_ert1', 'bar_qwe2_ert2', 'baz_qwe3_ert3'] 

L = [x.split('_') for x in l1]
a, b, c = L[0]
b = b.replace('1','')
c = c.replace('1','')
print (b, c)
qwe ert

out = [(y[0], f'{b}{x}', f'{c}{x}') for x, y in zip(array, L)]
print (out)
[('foo', 'qwe1', 'ert1'), ('bar', 'qwe3', 'ert3'), ('baz', 'qwe2', 'ert2')]

Or:

out = [(y[0], f'qwe{x}', f'ert{x}') for x, y in zip(array, L)]
print (out)
[('foo', 'qwe1', 'ert1'), ('bar', 'qwe3', 'ert3'), ('baz', 'qwe2', 'ert2')]

---

df = pd.DataFrame(out, columns=['Column1', 'Column2', 'Column3'])
print (df)
  Column1 Column2 Column3
0     foo    qwe1    ert1
1     bar    qwe3    ert3
2     baz    qwe2    ert2

Answered By - jezrael