我有一个包含18 000个唯一ID的列表。
ID是字母A, B, C, D
的串联。
我创建了一个代码,按ID[0:-1]
对ID进行分组,并给出重复ID的索引位置。
这种方法效果很好,但进展很长时间:110 secs
18 000 ID
左右。a = ['1CDABCABDA', '1CDABCABDB', '1CDABCABDD', '1BCABCCCAA', '1DDAABBBBA', '1BCABCCCAD']
startTime = time.time()
b = [i[0:-1] for i in a]
b = list(set(b))
result = range(len(b))
it = 0
for i in result:
result[i] = [b[i], []]
for j in xrange(len(a)):
if b[i] == a[j][0:-1]:
result[i][1].append(j)
endTime = time.time()
print endTime - startTime, 'secs !'
。
你有想法加快我的代码吗?
>>> [['1CDABCABD', [0, 1, 2]], ['1DDAABBBB', [4]], ['1BCABCCCA', [3, 5]]]
输出:
{{1}}
答案 0 :(得分:5)
这是python中groupby有效的方法:
from itertools import groupby
a = ['1CDABCABDA', '1CDABCABDB', '1CDABCABDD', '1BCABCCCAA', '1DDAABBBBA', '1BCABCCCAD']
key = lambda i: a[i][:-1]
indexes = sorted(range(len(a)), key=key)
result = [[x, list(y)] for x, y in groupby(indexes, key=key)]
输出:
[['1BCABCCCA', [3, 5]], ['1CDABCABD', [0, 1, 2]], ['1DDAABBBB', [4]]]
答案 1 :(得分:5)
作为解决此类问题的更多Pythonic方法,请使用collections.defaultdict
:
>>> from collections import defaultdict
>>> d=defaultdict(list)
>>> new=[i[:-1] for i in a]
>>> d=defaultdict(list)
>>> for i,j in enumerate(new):
... d[j].append(i)
...
>>> d
defaultdict(<type 'list'>, {'1CDABCABD': [0, 1, 2], '1DDAABBBB': [4], '1BCABCCCA': [3, 5]})
>>> d.items()
[('1CDABCABD', [0, 1, 2]), ('1DDAABBBB', [4]), ('1BCABCCCA', [3, 5])]
注意 defaultdict
是线性解决方案,效率高于itertools.groupby
和sorted
。
您也可以使用dict.setdefault
方法:
>>> d={}
>>> for i,j in enumerate(new):
... d.setdefault(j,[]).append(i)
...
>>> d
{'1CDABCABD': [0, 1, 2], '1DDAABBBB': [4], '1BCABCCCA': [3, 5]}
有关详细信息,请查看以下基准,标记为 ~4X 更快:
s1="""
from itertools import groupby
a = ['1CDABCABDA', '1CDABCABDB', '1CDABCABDD', '1BCABCCCAA', '1DDAABBBBA', '1BCABCCCAD']
key = lambda i: a[i][:-1]
indexes = sorted(range(len(a)), key=key)
result = [[x, list(y)] for x, y in groupby(indexes, key=key)]
"""
s2="""
a = ['1CDABCABDA', '1CDABCABDB', '1CDABCABDD', '1BCABCCCAA', '1DDAABBBBA', '1BCABCCCAD']
new=[i[:-1] for i in a]
d={}
for i,j in enumerate(new):
d.setdefault(j,[]).append(i)
d.items()
"""
print ' first: ' ,timeit(stmt=s1, number=100000)
print 'second : ',timeit(stmt=s2, number=100000)
结果:
first: 0.949549913406
second : 0.250894069672
答案 2 :(得分:2)
不使用其他模块的替代解决方案:
grouped = {}
for i, j in enumerate(a):
itm = grouped.get(j[0:-1], [])
itm.append(i)
grouped[j[0:-1]] = itm
print [[k, v] for k, v in grouped.items()] # [['1CDABCABD', [0, 1, 2]], ['1DDAABBBB', [4]], ['1BCABCCCA', [3, 5]]]
答案 3 :(得分:1)
你在找这个:
>>> d = {}
>>> for ind, elem in enumerate(a):
... d.setdefault(elem[0:-1], []).append(ind)
>>> print d
{'1CDABCABD': [0, 1, 2], '1DDAABBBB': [4], '1BCABCCCA': [3, 5]}
该解决方案与Kasra的优化代码非常相似,但工作速度稍快。不同之处在于切片完成的位置,但不确定为什么一个表现稍好于另一个:
s1 = """
a = ['1CDABCABDA', '1CDABCABDB', '1CDABCABDD', '1BCABCCCAA',
'1DDAABBBBA', '1BCABCCCAD']
d = {}
for ind, elem in enumerate(a):
d.setdefault(elem[0:-1], []).append(ind)
"""
s2="""
a = ['1CDABCABDA', '1CDABCABDB', '1CDABCABDD', '1BCABCCCAA', '1DDAABBBBA', '1BCABCCCAD']
new=[i[:-1] for i in a]
d={}
for i,j in enumerate(new):
d.setdefault(j,[]).append(i)
"""
print 'Kasra's time/my time: %s' % (str(timeit(stmt=s2, number=100000)/timeit(stmt=s1, number=100000))
Kasra's time/my time: 1.24058060531