skyfail skyfail - 10 months ago 37
Python Question

Iterate through a list of list in the form of a matrix but avoid a set of columns

So I have a list of list :-

a = [[1,2,3,4],[5,6,7,8],[4,5,6,7]]

I want to iterate through the contents and avoid a certain column. So my general for loop structure is as follows:

for i in range (0,len(a[0])):

Now as far as I know I need one more condition to avoid a certain column. How do I do that?

Another way this could be done is to delete the whole column and append it right to the end.

So after deleting the column:

a = [[1,2,4],[5,6,8],[4,5,7]]

After appending:

a = [[1,2,4,3],[5,6,8,7],[4,5,7,6]]

I could use numpy to do this. So I am deleting the desired column from numpy.
Here's a working code:

a = [[1,2,3,4],[5,6,7,8],[4,5,6,7]]
a_np= np.array(a)
a_col_data = a_np[:,2]
a1 = np.delete(a_np, 2 , axis = 1)

Now my a_col_data is
[3 7 6]

I cannot append in this format. I need it in this format

Then I could use the following code to append it as a last column in a:

np.append(a_np, a_col_data, axis=1)

Now issues with this approach, that I am facing:
How to convert a list
[3 7 6]
Considering I have a list of 150 columns and 3000 rows I can't do it manually.

Another issue is np.array converts the list into a pure matrix like structure removing the "," between elements. I want to get the list structure back.
For example:

a = [[1,2,4],[5,6,8],[4,5,7]]
a_np = [[1 2 4][5 6 8][4 5 7]]

How do I convert a_np to a again?


Ok I was just browsing stackoverflow and I got a solution. I could simply do the following to convert it back:


Answer Source

There are many ways to do this.

You could add an extra axis to a_col_data:

>>> a_np2 = np.append(a1, a_col_data[:, np.newaxis], axis=1)
>>> a_np2
array([[1, 2, 4, 3],
       [5, 6, 8, 7],
       [4, 5, 7, 6]])
>>> a_np2.tolist()
[[1, 2, 4, 3], [5, 6, 8, 7], [4, 5, 7, 6]]

Or you could avoid going the numpy route by using enumerate:

a = [[1,2,3,4],[5,6,7,8],[4,5,6,7]]
for i in a:
    for j, x in enumerate(i):
        if j == 2: