将字典项聚合为摘要结果

时间:2010-12-15 19:37:44

标签: python dictionary

我有一个像

这样的词典
    mydict={
        (a,1):0,
        (a,2):0,
        (a,3):0,
        (a,4):1,
        (a,5):2,
        (a,6):2,
        (a,7)=0,
        (a,8)=0,           
}

我想将其概括为

mysummarydict={
    (a,1,3):0,
    (a,4,4):1,
    (a,5,6):2,
    (a,7,8):0
    }

这些值来自关于不重叠但可能有间隙的间隔的数据集。第一个字典现在每个单点都有一个条目,我想得到第二个字典,其中包含那些具有共同值的相邻点的摘要。你能指导我在python 2.6中找到最好的解决方案吗? 感谢

3 个答案:

答案 0 :(得分:2)

from itertools import groupby
from operator import itemgetter

mydict={
        ('a', 1): 0,
        ('a', 2): 0,
        ('a', 3): 0,
        ('a', 4): 1,
        ('a', 5): 2,
        ('a', 6): 2,
        ('a', 7): 0,
        ('a', 8): 0,           
}

data = mydict.items()
data.sort()

def groupkey(item):
    return item[0][0], item[1]

result = {}
for v, group in groupby(data, key=groupkey):
    char, value = v
    nums = [item[0][1] for item in group]
    result[char, min(nums), max(nums)] = value

print result

结果:

{
 ('a', 1, 3): 0
 ('a', 4, 4): 1, 
 ('a', 5, 6): 2, 
 ('a', 7, 8): 0, 
}

答案 1 :(得分:0)

如果您将此数据存储在列表中,则会变得更加容易:

from itertools import groupby
from operator import itemgetter

mylist = [0, 0, 0, 1, 2, 2, 0, 0]

def interval(v):
    head = tail = next(v)
    for tail in v:
        pass

    return head[0] + 1, tail[0] + 1

print({interval(v): k for k, v in groupby(enumerate(mylist), key=itemgetter(1))})

{(5, 6): 2, (1, 3): 0, (7, 8): 0, (4, 4): 1}

答案 2 :(得分:0)

我找到了一种更短更快的方法:

from itertools import groupby
from operator import itemgetter
from time import clock

mydict={('a', 1): 0,
        ('a', 2): 0,
        ('a', 3): 0,
        ('a', 4): 1,
        ('a', 5): 2,
        ('a', 6): 2,
        ('a', 7): 0,
        ('a', 8): 0,
        }

A,B,C = [],[],[]

for i in xrange(1000):

    t0 = clock()
    data = mydict.items()
    data.sort()
    def groupkey(item):
        return item[0][0], item[1]
    result1 = {}
    for v, group in groupby(data, key=groupkey):
        char, value = v
        nums = [item[0][1] for item in group]
        result1[char, min(nums), max(nums)] = value
    A.append(clock()-t0)

    #----------------------------------------------------------------

    t0 = clock()
    data = [ [a,b,c] for ((a,b),c) in mydict.items()]
    data.sort()
    result2 = {}
    for (char,value),group in groupby(data, key=itemgetter(0,2)):
        nums = [item[1] for item in group]
        result2[char,nums[0],nums[-1]] = value
    B.append(clock()-t0)

    #-----------------------------------------------------------------

    t0 = clock()
    data = [ [a,b,c] for ((a,b),c) in mydict.items()]
    data.sort()
    result3 = {}
    for ((char,value),nums) in [ (cle,[item[1] for item in group]) for cle,group in groupby(data, key=itemgetter(0,2))]:
        result3[char,nums[0],nums[-1]] = value
    C.append(clock()-t0)

print 'result1==',result1
print 'result2==',result2
print 'result3==',result3
print 'result1==result2==result3==',result1==result2==result3
print id(result1)==id(result2),id(result2)==id(result3),id(result3)==id(result1)


print '{:.1%}.'.format(min(B)/min(A))
print '{:.1%}.'.format(min(C)/min(A))

结果:

result1 == {('a',5,6):2,('a',4,4):1,('a',7,8):0,('a',1, 3):0}

result2 == {('a',5,6):2,('a',4,4):1,('a',7,8):0,('a',1, 3):0}

result3 == {('a',5,6):2,('a',4,4):1,('a',7,8):0,('a',1, 3):0}

result1 == result2 == result3 == True

假错假

87.0%。

93.2%。