Python Pandas：更改行和标题

Question

我有以下数组：

[['fic10', {'bulle_naif': '55'}, {'bulle_bool': '52'}, {'bulle_opt': '39'}, {'selection': '45'}, {'insertion': '20'}, {'rapide': '60'}], ['fic100', {'bulle_naif': '5050'}, {'bulle_bool': '5050'}, {'bulle_opt': '4816'}, {'selection': '4950'}, {'insertion': '2221'}, {'rapide': '6697'}], ['fic1000', {'bulle_naif': '2623195'}, {'bulle_bool': '1789209'}, {'bulle_opt': '2618499'}, {'selection': '2620905'}, {'insertion': '1535788'}, {'rapide': '1323294'}], ['fic10000', {'bulle_naif': '4764881010'}, {'bulle_bool': '926117379'}, {'bulle_opt': '4764749559'}, {'selection': '4764783390'}, {'insertion': '900955079'}, {'rapide': '506697139'}]]

然后我将其转换为数据框。

这是我将数组转换为数据框时得到的结果：

my output

但我想像这样格式化我的数据框：

desired output

谁能帮帮我？

Answer 1

这是一段更简单易懂的代码，它可以帮助您解决问题。

import pandas as pd

array = [['fic10', {'bulle_naif': '55'}, {'bulle_bool': '52'}, {'bulle_opt': '39'}, {'selection': '45'}, {'insertion': '20'}, {'rapide': '60'}],
         ['fic100', {'bulle_naif': '5050'}, {'bulle_bool': '5050'}, {'bulle_opt': '4816'}, {'selection': '4950'}, {'insertion': '2221'}, {'rapide': '6697'}],
         ['fic1000', {'bulle_naif': '2623195'}, {'bulle_bool': '1789209'}, {'bulle_opt': '2618499'}, {'selection': '2620905'}, {'insertion': '1535788'}, {'rapide': '1323294'}],
         ['fic10000', {'bulle_naif': '4764881010'}, {'bulle_bool': '926117379'}, {'bulle_opt': '4764749559'}, {'selection': '4764783390'}, {'insertion': '900955079'}, {'rapide': '506697139'}]]

dt = pd.DataFrame()
for item in array:
    tmp = pd.DataFrame(index=[item[0]])
    for i in range(len(item)-1):
        if i != 0:
            for key, value in item[i].items():
                tmp[key] = value
    dt = dt.append(tmp)
print(dt)

输出：

          bulle_naif bulle_bool   bulle_opt   selection  insertion
fic10             55         52          39          45         20
fic100          5050       5050        4816        4950       2221
fic1000      2623195    1789209     2618499     2620905    1535788
fic10000  4764881010  926117379  4764749559  4764783390  900955079

Answer 2

让我们以pandas.DataFrame 更容易理解的方式呈现您的数据。

第一种方法：每列一个条目 key:list 的字典

此方法的目标是重新排列您的数据，使其看起来像一个字典，每列有一个条目，每列都有一个值列表：

# {'bulle_naif': ['55', '5050', '2623195', '4764881010'],
#  'bulle_bool': ['52', '5050', '1789209', '926117379'],
#  'bulle_opt': ['39', '4816', '2618499', '4764749559'],
#  'selection': ['45', '4950', '2620905', '4764783390'],
#  'insertion': ['20', '2221', '1535788', '900955079'],
#  'rapide': ['60', '6697', '1323294', '506697139'],
#  'name': ['fic10', 'fic100', 'fic1000', 'fic10000']}

这是进行该转换的代码：

import pandas as pd

raw_data = [['fic10', {'bulle_naif': '55'}, {'bulle_bool': '52'}, {'bulle_opt': '39'}, {'selection': '45'}, {'insertion': '20'}, {'rapide': '60'}], ['fic100', {'bulle_naif': '5050'}, {'bulle_bool': '5050'}, {'bulle_opt': '4816'}, {'selection': '4950'}, {'insertion': '2221'}, {'rapide': '6697'}], ['fic1000', {'bulle_naif': '2623195'}, {'bulle_bool': '1789209'}, {'bulle_opt': '2618499'}, {'selection': '2620905'}, {'insertion': '1535788'}, {'rapide': '1323294'}], ['fic10000', {'bulle_naif': '4764881010'}, {'bulle_bool': '926117379'}, {'bulle_opt': '4764749559'}, {'selection': '4764783390'}, {'insertion': '900955079'}, {'rapide': '506697139'}]]

cleaned_data = { k: [] for d in raw_data[0][1:] for k in d.keys() }
cleaned_data['name'] = []
for row in raw_data:
    cleaned_data['name'].append(row[0])
    for d in row[1:]:
        for k,v in d.items():
            cleaned_data[k].append(v)

print(cleaned_data)
# {'bulle_naif': ['55', '5050', '2623195', '4764881010'],
#  'bulle_bool': ['52', '5050', '1789209', '926117379'],
#  'bulle_opt': ['39', '4816', '2618499', '4764749559'],
#  'selection': ['45', '4950', '2620905', '4764783390'],
#  'insertion': ['20', '2221', '1535788', '900955079'],
#  'rapide': ['60', '6697', '1323294', '506697139'],
#  'name': ['fic10', 'fic100', 'fic1000', 'fic10000']}


# IMPORTANT NOTE
# This cleaning-up is a bit careless
# If a key is missing in one of the lists, the resulting data will be misaligned

# Making sure data is not misaligned:
assert(all(len(l) == len(cleaned_data['name']) for l in cleaned_data.values()))

good_dataframe = pd.DataFrame(cleaned_data).set_index('name')
print(good_dataframe)

#           bulle_naif bulle_bool   bulle_opt   selection  insertion     rapide
# name                                                                         
# fic10             55         52          39          45         20         60
# fic100          5050       5050        4816        4950       2221       6697
# fic1000      2623195    1789209     2618499     2620905    1535788    1323294
# fic10000  4764881010  926117379  4764749559  4764783390  900955079  506697139

第二种方法：一个没有键但按顺序排列的二维数组

如果您的数据已经排序，使得bulle_naif、bulle_opt 等在每一行的顺序都相同，那么您可以去掉所有的字典并直接为pandas.DataFrame 提供一个二维数组：

# assumes the rows of raw_data are all in the same order already
array_data = [[row[0]] + [v for d in row[1:] for v in d.values()] for row in raw_data]

print(array_data)
# [['fic10', '55', '52', '39', '45', '20', '60'],
#  ['fic100', '5050', '5050', '4816', '4950', '2221', '6697'],
#  ['fic1000', '2623195', '1789209', '2618499', '2620905', '1535788', '1323294'],
#  ['fic10000', '4764881010', '926117379', '4764749559', '4764783390', '900955079', '506697139']]

keys = ['name'] + [k for d in raw_data[0][1:] for k in d.keys()]
dataframe = pd.DataFrame(array_data, columns = keys).set_index('name')

print(dataframe)
#           bulle_naif bulle_bool   bulle_opt   selection  insertion     rapide
# name                                                                         
# fic10             55         52          39          45         20         60
# fic100          5050       5050        4816        4950       2221       6697
# fic1000      2623195    1789209     2618499     2620905    1535788    1323294
# fic10000  4764881010  926117379  4764749559  4764783390  900955079  506697139

如果您不知道所有行都已按相同顺序显示，则必须明确对它们进行排序以确保：

# I shuffled the entries in raw_data
raw_data = [
 ['fic10', {'bulle_bool': '52'}, {'bulle_naif': '55'}, {'selection': '45'}, {'insertion': '20'}, {'rapide': '60'}, {'bulle_opt': '39'}],
 ['fic100', {'bulle_opt': '4816'}, {'bulle_naif': '5050'}, {'insertion': '2221'}, {'selection': '4950'}, {'rapide': '6697'}, {'bulle_bool': '5050'}],
 ['fic1000', {'bulle_opt': '2618499'}, {'selection': '2620905'}, {'insertion': '1535788'}, {'bulle_bool': '1789209'}, {'rapide': '1323294'}, {'bulle_naif': '2623195'}],
 ['fic10000', {'selection': '4764783390'}, {'bulle_opt': '4764749559'}, {'bulle_bool': '926117379'}, {'bulle_naif': '4764881010'}, {'insertion': '900955079'}, {'rapide': '506697139'}]]

array_data = [[row[0]] + [v for d in sorted(row[1:], key=lambda d: next(iter(d.keys()))) for v in d.values()] for row in raw_data]

print(array_data)
# [['fic10', '52', '55', '39', '20', '60', '45'],
#  ['fic100', '5050', '5050', '4816', '2221', '6697', '4950'],
#  ['fic1000', '1789209', '2623195', '2618499', '1535788', '1323294', '2620905'],
#  ['fic10000', '926117379', '4764881010', '4764749559', '900955079', '506697139', '4764783390']]

Answer 3

创建数据框的问题在于数组的排序方式。我们需要重新排列数组以创建所需的数据框。

这段代码为数据框创建了两个格式正确的列表：

# create empty list for cols and rows
row_names = []
cols = []

# loop over the array
for i in arr:
    # add row name
    row_names.append((i[0]))
    # create dict for row values
    new_dict = {}
    # loop over row values
    for j in i[1:]:
        # add each value to new dict
        new_dict.update(j)
    # add row dict to columns
    cols.append(new_dict)

运行时我们会得到

row_names = ['fic10', 'fic100', 'fic1000', 'fic10000']
cols = [
    {'bulle_naif': '55', 'bulle_bool': '52', 'bulle_opt': '39', 'selection': '45', 'insertion': '20', 'rapide': '60'}, 
    {'bulle_naif': '5050', 'bulle_bool': '5050', 'bulle_opt': '4816', 'selection': '4950', 'insertion': '2221', 'rapide': '6697'}, 
    {'bulle_naif': '2623195', 'bulle_bool': '1789209', 'bulle_opt': '2618499', 'selection': '2620905', 'insertion': '1535788', 'rapide': '1323294'}, 
    {'bulle_naif': '4764881010', 'bulle_bool': '926117379', 'bulle_opt': '4764749559', 'selection': '4764783390', 'insertion': '900955079', 'rapide': '506697139'}
]

这可以很容易地转换为数据框：

df = pd.DataFrame(cols, index=row_names)

所以大家一起来：

import pandas as pd


arr = [['fic10', {'bulle_naif': '55'}, {'bulle_bool': '52'}, {'bulle_opt': '39'}, {'selection': '45'}, {'insertion': '20'}, {'rapide': '60'}], ['fic100', {'bulle_naif': '5050'}, {'bulle_bool': '5050'}, {'bulle_opt': '4816'}, {'selection': '4950'}, {'insertion': '2221'}, {'rapide': '6697'}], ['fic1000', {'bulle_naif': '2623195'}, {'bulle_bool': '1789209'}, {'bulle_opt': '2618499'}, {'selection': '2620905'}, {'insertion': '1535788'}, {'rapide': '1323294'}], ['fic10000', {'bulle_naif': '4764881010'}, {'bulle_bool': '926117379'}, {'bulle_opt': '4764749559'}, {'selection': '4764783390'}, {'insertion': '900955079'}, {'rapide': '506697139'}]]

# create empty list for cols and rows
row_names = []
cols = []

# loop over the array
for i in arr:
    # add row name
    row_names.append((i[0]))
    # create dict for row values
    new_dict = {}
    # loop over row values
    for j in i[1:]:
        # add each value to new dict
        new_dict.update(j)
    # add row dict to columns
    cols.append(new_dict)
# creta DataFrame
df = pd.DataFrame(cols, index=row_names)

输出：

          bulle_naif bulle_bool   bulle_opt   selection  insertion
fic10             55         52          39          45         20
fic100          5050       5050        4816        4950       2221
fic1000      2623195    1789209     2618499     2620905    1535788
fic10000  4764881010  926117379  4764749559  4764783390  900955079