Select column from a list

[hoffik]

Hello,

I'm quite new in Python and I have one question. I have a 2D matrix of
values stored in list (3 columns, many rows). I wonder if I can select one
column without having to go through the list with 'for' command.

For example I have list called 'values'.
When I write 'values[0]' or 'values[0][:]' I'll get the first row.
But when I write 'values[:][0]' I won't get the first column, but the first
row again! I can't see why.

Thanks
Hoffik

 

[Bruno Desthuilliers]

hoffik a écrit :

Hello,

I'm quite new in Python and I have one question. I have a 2D matrix of
values stored in list (3 columns, many rows). I wonder if I can select one
column without having to go through the list with 'for' command.

Lists don't have columns. What you have is a list of lists (and FWIW, in
your case, it should be a list of tuples since obviously these are
fixed-size items where position is meaningfull).

For example I have list called 'values'.
When I write 'values[0]' or 'values[0][:]' I'll get the first row.

Not exactly. In the first case, you have a reference to values[0], in
the second case a _copy_ of values[0]

>>> alist = [[1, 2, 3], [4, 5, 6]]
>>> ref = alist[0]
>>> copy = alist[0][:]
>>> ref is alist[0] True
>>> copy is alist[0] False
>>> copy[0] = 999
>>> copy [999, 2, 3]
>>> alist [[1, 2, 3], [4, 5, 6]]
>>> ref [1, 2, 3]
>>> ref[0] = 888
>>> alist [[888, 2, 3], [4, 5, 6]]
>>>

But when I write 'values[:][0]' I won't get the first column, but the first
row again! I can't see why.

alist[:] returns a copy the whole list. As I said, lists don't have
"columns".

So yes, given your data structure, you do have to iterate over the outer
list to collect items at specific positions of the inner lists (or tuples).

Depending on the size of your dataset, the frequency of such operation
in your code, and required perfs, you can either

* just do the simplest thing, ie:

first_column_items = [row[0] for row in values]

* try to find a more suited datastructure

* of just use the most appropriate tool for set operations, which is a
relational database. FWIW, sqlite doesn't require any server setup, and
can even be used with in-memory databases if you don't need persistance.

HTH

 

[Piet van Oostrum]

h> Hello,

h> I'm quite new in Python and I have one question. I have a 2D matrix of
h> values stored in list (3 columns, many rows). I wonder if I can select one
h> column without having to go through the list with 'for' command.

h> For example I have list called 'values'.
h> When I write 'values[0]' or 'values[0][:]' I'll get the first row.
h> But when I write 'values[:][0]' I won't get the first column, but the first
h> row again! I can't see why.

If you work with matrices, numpy may be interesting for you. It does
have the required functionality: M[:,0] will give you the first column.

 

[Alan G Isaac]

I'm quite new in Python and I have one question. I have a 2D matrix of
values stored in list (3 columns, many rows). I wonder if I can select one
column without having to go through the list with 'for' command.

Not quite what you asked but ...

rows = [(1, 5, 9), (2, 6, 10), (3, 7, 11), (4, 8, 12)]
cols = map(None,*rows)
cols

[(1, 2, 3, 4), (5, 6, 7, 8), (9, 10, 11, 12)]

Now you can have any "column" you want.

Alan Isaac

PS You can also use imap if you prefer not to
create the list.