Python 菜鸡请教

[wsds]

有这么一堆数据：

#test.log
a, 1.324171
b, 0.000126
c, 1.970941
a, 1.469649
b, 0.000124
c, 0.512929
a, 1.290920
b, 0.000118
c, 0.259524
a, 0.495958
b, 0.000123
c, 0.910949
a, 1.268038
b, 0.000118
c, 1.016419
a, 1.856081
b, 0.000120
c, 1.400075
a, 1.314131
b, 0.000140

想要用 python 把左边的 key 一样的合并，但 value 要取它所有的和，还有平均值

搞了半天，发现搞不定，也是尴尬,以下还是个半成品，搞不下去了，报错，求大神指点一些简单方法

def two(file):

	arr = []
	with open(file, "r", encoding="utf-8") as f:
		for i in f.readlines():
			a = i.replace("\n", '').strip()
			if a.split(",")[0] not in arr:
				arr.append(a.split(",")[0])
	ser = -1
	while True:
		ser += 1
		try:
			if a.split(",")[ser] == arr[ser]:
				print(a.split(",")[ser])
		except IndexError:
			print("end!")
			break


two("test.log")

[linxzh1989]

pandas groupby sum?

[aborigine]

了解一下 pandas ？导入生成个 dataframe 就行了

[wsds]

@aborigine 要用纯 python 处理，该怎么弄？

[wsds]

@aborigine 我的意思是，不要用这种一步到位的库

[deepreader]

Dictionary 会用么？

[wsds]

@deepreader 抱歉，字典我知道，但以我的水平真心实现不了

[mentalkiller]

@wsds #6 dict 是 python 内置的数据结构啊，不需要你自己实现

[wsds]

@mentalkiller 我的意思不是实现字典的功能，我的意思是用字典，我实现不了我的需求

[hahastudio]

https://gist.github.com/hahastudio/d096995dc535abcad6046d79e238add4

[lixm]

就这需求还要上 pandas ？

https://gist.github.com/comynli/233412cd89bf671d7f9d754a3137be4f

[ybping]

dict 了解一下

[scriptB0y]

In [18]: data = """
...: a, 1.324171
...: b, 0.000126
...: c, 1.970941
...: a, 1.469649
...: b, 0.000124
...: c, 0.512929
...: a, 1.290920
...: b, 0.000118
...: c, 0.259524
...: a, 0.495958
...: b, 0.000123
...: c, 0.910949
...: a, 1.268038
...: b, 0.000118
...: c, 1.016419
...: a, 1.856081
...: b, 0.000120
...: c, 1.400075
...: a, 1.314131
...: b, 0.000140
...: """

In [19]: result = {}
...: for line in data.splitlines():
...: if not line: continue
...: key, value = line.split(",")
...: result.setdefault(key, []).append(float(value))
...:

In [20]: for key, values in result.items():
...: print(f"{key}: avg: {sum(values) / len(values)}, sum: {sum(values)}")
...:
a: avg: 1.2884211428571428, sum: 9.018948
b: avg: 0.00012414285714285714, sum: 0.000869
c: avg: 1.0118061666666667, sum: 6.070837

https://gist.github.com/laixintao/f4a186cea6c28fcf3dc696100458c410

[wsds]

@hahastudio
@lixm
@scriptB0y
非常感谢各位！学习了

[wplct]

def two(file):
data = {}
with open(file, "r") as f:
while True:
s = f.readline()
if s is None or not s:
break
print(s.split(', '))
k, v = s.split(', ')
v = float(v)
if k not in data:
data[k] = {
'num': 1,
'sum': v,
'avg': v
}
else:
data[k]['num'] += 1
data[k]['sum'] += v
data[k]['avg'] = data[k]['sum'] / data[k]['num']
print(data)
two('test.txt')

[wsds]

@wplct 感谢

[araraloren]

一行脚本拯救你

perl -nE 'state %z; my @z = split(", "); $z{@z[0]} += @z[1]; END { say for %z; } ' < test.log
a
9.018948
b
0.000869
c
6.070837

perl6 -ne 'state %z; given .split(", ") { %z{.[0]} += .[1].Rat; }; END { say %z; }' < test.log
{a => 9.018948, b => 0.000869, c => 6.070837}

[bufpay]

其实一个 for 循环就可以了

[E1n]

@araraloren perl 好用，用 awk 能实现吗。。

[araraloren]

@E1n awk 肯定能实现，不过我只懂基本的 awk 脚本

[imagechans]

def stand(file):
datas = [str(line).replace("\n","").strip().split(',')[1] for line in open(file)]
s = sum([float(d) for d in datas])
m = s / len(datas)
print(s,m)


stand("test.log")

[imagechans]

@imagechans 这是按照我自己的习惯写的

[billgreen1]

< test.log | awk -F"," '{total[$1]+=$2; occurence[$1]+=1}END{for (key in total) printf("%s\t %s \t %s\n", key, total[key], total[key]/occurence[key])}'

[slimbloody]

default_dict

[arthasgxy]

@lixm
http://chuantu.biz/t6/328/1529036276x-1404792211.jpg

求助，不知道为什么我这里运行会这样

[arthasgxy]

https://gist.github.com/arthasgxy/9d1f8aae0c1e90dde5ec5e44032be4f5

字典用的少，感觉只能写出来这种又蠢又长的

[wsds]

@bufpay 发现越是看上去简单的功能，实现起来越是困难

[xpresslink]

>>> data = """a, 1.324171
b, 0.000126
c, 1.970941
a, 1.469649
b, 0.000124
c, 0.512929
a, 1.290920
b, 0.000118
c, 0.259524
a, 0.495958
b, 0.000123
c, 0.910949
a, 1.268038
b, 0.000118
c, 1.016419
a, 1.856081
b, 0.000120
c, 1.400075
a, 1.314131
b, 0.000140"""
>>> import csv
>>> from itertools import groupby
>>> from operator import itemgetter as ig
>>> {k:sum(map(lambda x:float(ig(1)(x)), v)) for k, v in groupby(sorted(csv.reader(iter(data.splitlines())), key=ig(0)), key=ig(0))}
{'a': 9.018948, 'b': 0.000869, 'c': 6.070837}
>>>

[Xiaobaixiao]

from collections import defaultdict

logData='''a, 1.324171
b, 0.000126
c, 1.970941
a, 1.469649
b, 0.000124
c, 0.512929
a, 1.290920
b, 0.000118
c, 0.259524
a, 0.495958
b, 0.000123
c, 0.910949
a, 1.268038
b, 0.000118
c, 1.016419
a, 1.856081
b, 0.000120
c, 1.400075
a, 1.314131
b, 0.000140'''

def solve(logData):
logList=[]
logDict=defaultdict(int)
for line in logData.splitlines():
newList = line.split(',')
k=newList[0]
v=float(newList[1])
logList.append(k)
logDict[k]+=v
for k,v in logDict.items():
avg = v/logList.count(k)
print("{0} 总和：{1} , 平均值：{2}".format(k,v,avg))


>>> solve(logData)
a 总和：9.018948 , 平均值：1.2884211428571428
b 总和：0.000869 , 平均值：0.00012414285714285714
c 总和：6.070837 , 平均值：1.0118061666666667

[zhang0320]

with open('data.txt', 'r') as f:
data = f.read()
dict_data = {}
for i in data.split('\n'):
if i.split(',')[0] not in dict_data:
dict_data[i.split(',')[0]]=float(i.split(',')[1])
dict_data[i.split(',')[0]]=dict_data[i.split(',')[0]]+float(i.split(',')[1])

print(dict_data)



我是个新手菜鸟，不知道这种想法对不对。。。

[zhang0320]

with open('data.txt', 'r') as f:
data = f.read()
dict_data = {}
for i in data.split('\n'):
if i.split(',')[0] not in dict_data:
dict_data[i.split(',')[0]] = float(i.split(',')[1])
else:
dict_data[i.split(',')[0]] = dict_data[i.split(',')[0]] + float(i.split(',')[1])

print(dict_data)


不好意思 忘了个 else：

[Alexhex]

拷到 Excel 里一个数据透视表搞定美滋滋。

[ful1v1dcker]

什么鬼，V 站 markdown 失效了？全是代码坨啊。。。

[reself]

@araraloren 哈哈 perl 自带代码混淆

[araraloren]

@ful1v1dcker 本来就是这样，只支持楼主的。。不然还需要什么 chrome markdown 插件（外面那个帖子）

[JCZ2MkKb5S8ZX9pq]

d = {}
逐行
k, v = (i.strip() for i in text.split(','))
d.setdefault(k, []) # 建个列表
d[k].append(float(v))

这样直观一点，然后求值啥的慢慢折腾呗。

[lixm]

@arthasgxy 因为是 yield， 所以是一个生成器

[gnozix]

我意思一下：
```python
temp = {}
for i, j in dict or tuple:
if i in temp:
temp[i] = float(j)
else:
temp[i] += float(j)
```

[ful1v1dcker]

@araraloren 噢好吧，果然没用最垃圾只有更垃圾

[yaorc]

def pivot_table():
with open('data.txt', 'r') as f:
all_data = f.readlines()[1:]
keys = []
result = {}
count = {}

for data in all_data:
content = data.split(',')
key = content[0]
value = float(content[1].strip())

# 添加键
if key not in keys:
keys.append(key)
result[key] = value
count[key] = 1
else:
result[key] += value
count[key] += 1

print('元素：', keys)
print('元素个数：', count)
print('和：', result)

print('\n 统计信息（元素，和，平均值）：')
for k, v in result.items():
avg = v / count[k]
print(k, v, avg)


----------输出-----------
元素： ['a', 'b', 'c']
元素个数： {'a': 7, 'b': 7, 'c': 6}
和： {'a': 9.018948, 'b': 0.000869, 'c': 6.070837}

统计信息（元素，和，平均值）：
a 9.018948 1.2884211428571428
b 0.000869 0.00012414285714285714
c 6.070837 1.0118061666666667

[gpj22pYlv2qYiZ8U]

def two(file):

num_dict = {}
with open(file, "r", encoding="utf-8") as f:
for i in f.readlines():
a = i.replace("\n", '').strip()
line_list = a.split(",")
if line_list[0] not in num_dict:
num_dict[line_list[0]] = [line_list[1], 1]
else:
num_dict[line_list[0]] = [float(num_dict[line_list[0]][0]) + float(line_list[1]), int(num_dict[line_list[0]][1]) + 1]
for x in num_dict:
num_dict[x].append(num_dict[x][0] / num_dict[x][1])
print(num_dict)


two("/Users/yourname/program/test/test.log")

[UnluckyNinja]

def text = '''a, 1.324171
b, 0.000126
c, 1.970941
a, 1.469649
b, 0.000124
c, 0.512929
a, 1.290920
b, 0.000118
c, 0.259524
a, 0.495958
b, 0.000123
c, 0.910949
a, 1.268038
b, 0.000118
c, 1.016419
a, 1.856081
b, 0.000120
c, 1.400075
a, 1.314131
b, 0.000140
'''

// file.readLines().collect{
text.readLines().collect{
it.split(',')*.trim()
}.groupBy{
it[0]
}.collectEntries{k, vList ->
[(k): [sum: def sum = vList.sum{ it[1] as BigDecimal }, average: sum / vList.size()]]
}

/* result: [a:[sum:9.018948, average:1.2884211429], b:[sum:0.000869, average:0.0001241429], c:[sum:6.070837, average:1.0118061667]]

groovy 写的，groovy 有的 python 肯定有，语法方法名啥的改一下应该就差不多了 */

[Binb]

#!/usr/bin/env python
# -*- coding: utf-8 -*-

a = {}
with open('test.log','r') as f:
for i in f:
j = i.split(',')
a.setdefault(j[0],[]).append(float(j[1]))

result = {}
for k,v in a.items():
s = sum(v)
result.setdefault(k,[]).append(s)
result[k].append(s/len(v))

print result
# {'a': [9.018948, 1.2884211428571428], 'c': [6.070837, 1.0118061666666667], 'b': [0.000869, 0.00012414285714285714]}

[NICCEEEE]

import collections

d = """a, 1.324171
b, 0.000126
c, 1.970941
a, 1.469649
b, 0.000124
c, 0.512929
a, 1.290920
b, 0.000118
c, 0.259524
a, 0.495958
b, 0.000123
c, 0.910949
a, 1.268038
b, 0.000118
c, 1.016419
a, 1.856081
b, 0.000120
c, 1.400075
a, 1.314131
b, 0.000140"""

L = [(i[0], i[3:]) for i in d.split('\n')]
data_dict = collections.defaultdict(int)

for i, j in L:
data_dict[i] += float(j)

print(data_dict)

[yangxiaoyong]

讲下原理的东西，map reduce 可以了解一下，首先把数据分组归类
map (lambda x: { value: x.key, key: x.key, count: 1})
按上面的把数据按 key 分组放好
然后执行归约函数，将数据集合归约为一个最终结果
reduce(lambda acc, curr: merge(acc,curr), mapdata ）

merge 根据 key 将相同 key 的数值相加得到总合，count 相加得到次数，总和除以次数可以得平均值

最后的结果应该是 {a: { value，count，avg}}

手机码字，凑合看吧

[bugcoder]

没人贴 pandas 的，我就献个丑吧：
···
import pandas as pd

data_file = 'data.txt'
data_df = pd.read_csv(data_file, comment='#', names=['key', 'value'])

sums = data_df.groupby('key').sum()
means = data_df.groupby('key').mean()
···

[mingyun]

@bugcoder 这个简单

[biglazycat]

convert_list = {}
for line in open('test.log'):
k, v = line.split(',')
convert_list.setdefault(k,[]).append(float(v.strip()))

# print(convert_list)
for k, v in convert_list.items():
total_sum = sum(v)
avg = total_sum / len(v)
print(total_sum)
print(avg)