SortingHat - 3 years ago 290
Python Question

# Numpy: Insert value into different column for each row

I'm trying to insert into a numpy matrix given a mask that defines a single cell per row. Effectively, it's inserting a value into each row but with a different column. I've tried to use

`np.insert()`
without success:

``````>>> x
array([[False, False,  True, False, False],
[False, False,  True, False, False],
[False, False,  True, False, False],
[False, False,  True, False, False],
[False, False,  True, False, False]], dtype=bool)
>>> y = np.arange(25).reshape(5,5)
>>> y
array([[ 0,  1,  2,  3,  4],
[ 5,  6,  7,  8,  9],
[10, 11, 12, 13, 14],
[15, 16, 17, 18, 19],
[20, 21, 22, 23, 24]])
>>> np.insert(y, np.where(x)[1], 99, axis=1)
array([[ 0,  1, 99, 99, 99, 99, 99,  2,  3,  4],
[ 5,  6, 99, 99, 99, 99, 99,  7,  8,  9],
[10, 11, 99, 99, 99, 99, 99, 12, 13, 14],
[15, 16, 99, 99, 99, 99, 99, 17, 18, 19],
[20, 21, 99, 99, 99, 99, 99, 22, 23, 24]])
``````

Anytime I try and insert based on the
`x`
mask, it ends up duplicating values.

Also as noted, the mask may potentially be setup in a way that it's not a simple column. For example:

``````>>> x = np.zeros((5, 5), dtype=bool)
>>> x[1:, 2] = True
>>> x[0, 1] = True
>>> x
array([[False,  True, False, False, False],
[False, False,  True, False, False],
[False, False,  True, False, False],
[False, False,  True, False, False],
[False, False,  True, False, False]], dtype=bool)
``````

Which then means I can't simply specify a particular column as the index to insert at:

``````>>> np.insert(y, 2, [99, 99, 99, 99, 99], axis=1)
array([[ 0,  1, 99,  2,  3,  4],
[ 5,  6, 99,  7,  8,  9],
[10, 11, 99, 12, 13, 14],
[15, 16, 99, 17, 18, 19],
[20, 21, 99, 22, 23, 24]])
``````

The desired output would be:

``````array([[ 0,  99, 1,  2,  3,  4],
[ 5,  6, 99,  7,  8,  9],
[10, 11, 99, 12, 13, 14],
[15, 16, 99, 17, 18, 19],
[20, 21, 99, 22, 23, 24]])
``````

Any help would be greatly appreciated!

Approach #1 : Here's one way with `boolean-indexing` -

``````def insert_one_per_row(arr, mask, putval):
return out
``````

Sample run -

``````In [88]: y
Out[88]:
array([[ 0,  1,  2,  3,  4],
[ 5,  6,  7,  8,  9],
[10, 11, 12, 13, 14],
[15, 16, 17, 18, 19],
[20, 21, 22, 23, 24]])

In [89]: x
Out[89]:
array([[False,  True, False, False, False],
[False, False,  True, False, False],
[False, False, False, False,  True],
[ True, False, False, False, False],
[False, False,  True, False, False]], dtype=bool)

In [90]: insert_one_per_row(y, x, putval=99)
Out[90]:
array([[ 0, 99,  1,  2,  3,  4],
[ 5,  6, 99,  7,  8,  9],
[10, 11, 12, 13, 99, 14],
[99, 15, 16, 17, 18, 19],
[20, 21, 99, 22, 23, 24]])
``````

We can also assign different values per row -

``````In [91]: insert_one_per_row(y, x, putval=[-1,-2,-3,-4,-5])
Out[91]:
array([[ 0, -1,  1,  2,  3,  4],
[ 5,  6, -2,  7,  8,  9],
[10, 11, 12, 13, -3, 14],
[-4, 15, 16, 17, 18, 19],
[20, 21, -5, 22, 23, 24]])
``````

Approach #2 : We will get the flattened `True` places on the mask and insert the new values with `np.insert` on a flattened version of the input array at those places, like so -

``````def insert_one_per_row_v2(arr, mask, putval):