Using reverse iteration to clean up a list

[tkpmep@hotmail.com]

I have list of lists of the following form

L=[['A', 100], ['B', 300], ['A', 400], ['B', -100]]

I want to aggregate these lists, i.e. to reduce L to
L=[['A', 500], ['B', 200]] #500 = 100+400, 200=300-100

Here's how I have done it:
L.sort()
for i in range(len(L),0,-1):
if L[i-1][0]=L[i][0]:
L[i-1][2] += L[i][2]
del L[i]

Is there a simpler way to do this using the new reverse iteration
facility in Python 2.4?

Thomas Philips

[Paul Rubin]

Re: Using reverse iteration to clean up a list

tkpmep@hotmail. com writes:[color=blue]
> I have list of lists of the following form
>
> L=[['A', 100], ['B', 300], ['A', 400], ['B', -100]]
>
> I want to aggregate these lists, i.e. to reduce L to
> L=[['A', 500], ['B', 200]] #500 = 100+400, 200=300-100[/color]

How about:

v = {}
for name,val in L:
v[name] = v.get(name, 0) + val
L = v.items()

[John Machin]

Re: Using reverse iteration to clean up a list


Paul Rubin wrote:[color=blue]
> tkpmep@hotmail. com writes:[color=green]
> > I have list of lists of the following form
> >
> > L=[['A', 100], ['B', 300], ['A', 400], ['B', -100]]
> >
> > I want to aggregate these lists, i.e. to reduce L to
> > L=[['A', 500], ['B', 200]] #500 = 100+400, 200=300-100[/color]
>
> How about:
>
> v = {}
> for name,val in L:
> v[name] = v.get(name, 0) + val
> L = v.items()[/color]

Followed by L.sort() if the OP really needs his list sorted.
Alternatively, he may prefer to leave it in a dict; in fact, if he had
been using a dict all along and maintaining it on the fly by "v[name]
= v.get(name, 0) + val", he wouldn't need to batch-fix his data
structure now.

tkpmep, how did you get this list of lists with duplicate keys and
additive values? What do you want to do with it next?

[Luis M. Gonzalez]

Re: Using reverse iteration to clean up a list

Well, I'm not sure if this is what you want, but you could use a
dictionary:
[color=blue][color=green][color=darkred]
>>> d={}
>>> for i,e in L:[/color][/color][/color]
if d.has_key(i):
d[i] += e
else:
d[i] = e

[color=blue][color=green][color=darkred]
>>> d[/color][/color][/color]
{'A': 500, 'B': 200}[color=blue][color=green][color=darkred]
>>>[/color][/color][/color]

[Michael Hoffman]

Re: Using reverse iteration to clean up a list

tkpmep@hotmail. com wrote:
[color=blue]
> L=[['A', 100], ['B', 300], ['A', 400], ['B', -100]]
>
> I want to aggregate these lists, i.e. to reduce L to
> L=[['A', 500], ['B', 200]] #500 = 100+400, 200=300-100[/color]

"""
from itertools import groupby
from operator import itemgetter

[(key, sum(item[1] for item in sublist))
for key, sublist
in groupby(sorted( L, key=itemgetter( 0)), itemgetter(0))]
"""

OK, strictly speaking that will get you a list of tuples
instead of a list of lists, but that's easy enough to fix
if you insist on a list of lists. Tuples are generally used
for heterogeneous fixed-length sequences.
--
Michael Hoffman

[tkpmep@hotmail.com]

Re: Using reverse iteration to clean up a list

Thank you all so much for the generous dollop of help: the dictionary
suggestion is particularly helpful. The problem arises as follows: A
software application stores the securities held in a portfolio in a
..csv file, one row per security, with three colulmns.

The first has a security identifier or descriptor (such as a ticker)
the second has a single letter that identifies the type of the
identifier (T for ticker, C for cusip etc.) and the third has the
number of shares. A typical line looks like this:

IBM, T, 500

I need to read in one or more portfolios and aggregate their holdings.
To do so, I read in the portfolios using the csv package, convert each
line to a list and then append it to a list of lists. Eventually the
list of lists contains all the securities, and can then be sorted and
aggregated.

I suppose I could convert it instead to a dictionary, and the most
natural key would be the first two items, i.e. a portfolio containing
just 500 shares of IBM ought to be represented as
{("IBM", "T") : 500 }

How can I translate the data I read in using csv.reader into a
dictionary?

Thomas Philips

[Michael Hoffman]

Re: Using reverse iteration to clean up a list

tkpmep@hotmail. com wrote:[color=blue]
> Thank you all so much for the generous dollop of help: the dictionary
> suggestion is particularly helpful. The problem arises as follows: A
> software application stores the securities held in a portfolio in a
> .csv file, one row per security, with three colulmns.
>
> The first has a security identifier or descriptor (such as a ticker)
> the second has a single letter that identifies the type of the
> identifier (T for ticker, C for cusip etc.) and the third has the
> number of shares. A typical line looks like this:
>
> IBM, T, 500
>
> I need to read in one or more portfolios and aggregate their holdings.
> To do so, I read in the portfolios using the csv package, convert each
> line to a list and then append it to a list of lists. Eventually the
> list of lists contains all the securities, and can then be sorted and
> aggregated.
>
> I suppose I could convert it instead to a dictionary, and the most
> natural key would be the first two items, i.e. a portfolio containing
> just 500 shares of IBM ought to be represented as
> {("IBM", "T") : 500 }
>
> How can I translate the data I read in using csv.reader into a
> dictionary?[/color]

portfolio = {}

for row in csv.reader(infi le):
key = tuple(row[:2])

portfolio[key] = portfolio.get(k ey, 0) + int(row[2])

You could also do a groupby solution with
itemgetter(slic e(0, 2))--thanks to Steven Bethard for recently
pointing out the possibility here of doing that. I'd go with the
dict for this application though.
--
Michael Hoffman