CODEWITHSUNDEEP
pythonnumbersiteration
Adam Magyar
Adam Magyar

Reputation: 1423

Grouping / clustering numbers in Python

I've googled, I've tested, and this has me at my wits end. I have a list of numbers I need to group by similarity. For instance, in a list of [1, 6, 9, 100, 102, 105, 109, 134, 139], 1 6 9 would be put into a list, 100, 102, 105, and 109 would be put into a list, and 134 and 139. I'm terrible at math, and I've tried and tried this, but I can't get it to work. To be explicit as possible, I wish to group numbers that are within 10 values away from one another. Can anyone help? Thanks.

Upvotes: 22

Views: 25861

Answers (3)

abarnert
abarnert

Reputation: 366083

First, you can easily convert any sequence into a sequence of pairs of adjacent items. Just tee it, shift it forward, and zip the unshifted and unshifted copies. The only trick is that you need to start with either (<something>, 1) or (139, <something>), because in this case we want not each pair of elements, but a pair for each element:

def pairify(it):
    it0, it1 = itertools.tee(it, 2)
    first = next(it0)
    return zip(itertools.chain([first, first], it0), it1)

(This isn't the simplest way to write it, but I think this may be the way that's most readable to people who aren't familiar with itertools.)

>>> a = [1, 6, 9, 100, 102, 105, 109, 134, 139]
>>> list(pairify(a))
[(1, 1), (1, 6), (6, 9), (9, 100), (100, 102), (102, 105), (105, 109), (109, 134), (134, 139)]

Then, with a slightly more complicated version of Ned Batchelder's key, you can just use groupby.

However, I think in this case this will end up being more complicated than an explicit generator that does the same thing.

def cluster(sequence, maxgap):
    batch = []
    for prev, val in pairify(sequence):
        if val - prev >= maxgap:
            yield batch
            batch = []
        else:
            batch.append(val)
    if batch:
        yield batch

Upvotes: 3

Ned Batchelder
Ned Batchelder

Reputation: 375992

This will find the groups:

nums = [1, 6, 9, 100, 102, 105, 109, 134, 139]
for k, g in itertools.groupby(nums, key=lambda n: n//10):
    print k, list(g)

0 [1, 6, 9]
10 [100, 102, 105, 109]
13 [134, 139]

Note that if nums isn't actually sorted as your sample shows, you'll need to sort it first.

Upvotes: 8

Raymond Hettinger
Raymond Hettinger

Reputation: 226674

There are many ways to do cluster analysis. One simple approach is to look at the gap size between successive data elements:

def cluster(data, maxgap):
    '''Arrange data into groups where successive elements
       differ by no more than *maxgap*

        >>> cluster([1, 6, 9, 100, 102, 105, 109, 134, 139], maxgap=10)
        [[1, 6, 9], [100, 102, 105, 109], [134, 139]]

        >>> cluster([1, 6, 9, 99, 100, 102, 105, 134, 139, 141], maxgap=10)
        [[1, 6, 9], [99, 100, 102, 105], [134, 139, 141]]

    '''
    data.sort()
    groups = [[data[0]]]
    for x in data[1:]:
        if abs(x - groups[-1][-1]) <= maxgap:
            groups[-1].append(x)
        else:
            groups.append([x])
    return groups

if __name__ == '__main__':
    import doctest
    print(doctest.testmod())

Upvotes: 45

Related Questions