在没有任何聚合的情况下在数据框中传播特定列？

Question

这是我的玩具 df:

{'id': {0: 1089577, 1: 1089577, 2: 1089577, 3: 1089577, 4: 1089577},
 'title': {0: 'Hungarian Goulash Stew',
  1: 'Hungarian Goulash Stew',
  2: 'Hungarian Goulash Stew',
  3: 'Hungarian Goulash Stew',
  4: 'Hungarian Goulash Stew'},
 'readyInMinutes': {0: 120, 1: 120, 2: 120, 3: 120, 4: 120},
 'nutrients.amount': {0: 323.18, 1: 15.14, 2: 4.43, 3: 38.95, 4: 34.64},
 'nutrients.name': {0: 'Calories',
  1: 'Fat',
  2: 'Saturated Fat',
  3: 'Carbohydrates',
  4: 'Net Carbohydrates'},
 'nutrients.percentOfDailyNeeds': {0: 16.16,
  1: 23.3,
  2: 27.69,
  3: 12.98,
  4: 12.6},
 'nutrients.title': {0: 'Calories',
  1: 'Fat',
  2: 'Saturated Fat',
  3: 'Carbohydrates',
  4: 'Net Carbohydrates'},
 'nutrients.unit': {0: 'kcal', 1: 'g', 2: 'g', 3: 'g', 4: 'g'}}

我想将 nutrients.title 展开为专栏。 Sp 我将得到 Fat, Saturated Fat ... 列及其相应的值，没有任何聚合。

什么函数可以在没有任何聚合的情况下做到这一点？只是“重塑”而已。

我希望它是：

我怎样才能这样“传播”它？

Answer 1

试试 pivot_table:

# Rename Columns
df.columns = df.columns.map(lambda x: f".{x.split('.')[-1]}" if '.' in x else x)

# Create Pivot Table
df = df.pivot_table(
    index=['id', 'title', 'readyInMinutes'],
    columns=['.title'],
    values=['.amount',
            '.percentOfDailyNeeds',
            '.unit'],
    aggfunc='first'
).reset_index() \
    .swaplevel(0, 1, axis=1)

# Re-Order Columns So that nutrients.title are grouped
df = df.reindex(sorted(df.columns), axis=1)

# Reduce Levels by join
df.columns = df.columns.map(''.join)

print(df.to_string(index=False))

输出：

     id  readyInMinutes                  title  Calories.amount  Calories.percentOfDailyNeeds Calories.unit  Carbohydrates.amount  Carbohydrates.percentOfDailyNeeds Carbohydrates.unit  Fat.amount  Fat.percentOfDailyNeeds Fat.unit  Net Carbohydrates.amount  Net Carbohydrates.percentOfDailyNeeds Net Carbohydrates.unit  Saturated Fat.amount  Saturated Fat.percentOfDailyNeeds Saturated Fat.unit
1089577             120 Hungarian Goulash Stew           323.18                         16.16          kcal                 38.95                              12.98                  g       15.14                     23.3        g                     34.64                                   12.6                      g                  4.43                              27.69                  g

具有删节输出的步骤

更改列名称：

print(df.columns.values)
# ['id' 'title' 'readyInMinutes' 'nutrients.amount' 'nutrients.name'
#  'nutrients.percentOfDailyNeeds' 'nutrients.title' 'nutrients.unit']
print(df.columns.map(lambda x: f".{x.split('.')[-1]}" if '.' in x else x).values)
# ['id' 'title' 'readyInMinutes' '.amount' '.name' '.percentOfDailyNeeds'
#  '.title' '.unit']

使用单个 header 列对多个值列进行透视以创建 multi-level 列索引：

print(df.pivot_table(
    index=['id', 'title', 'readyInMinutes'],
    columns=['.title'],
    values=['.amount',
            '.percentOfDailyNeeds',
            '.unit'],
    aggfunc='first'
).to_string())

                                               .amount
.title                                        Calories Carbohydrates    Fat Net Carbohydrates Saturated Fat
id      title                  readyInMinutes
1089577 Hungarian Goulash Stew 120              323.18         38.95  15.14             34.64          4.43

修复索引和交换级别，使标签位于顶部（Calories、Carbohydrates 等） .reset_index().swaplevel(0, 1, axis=1)

.title                                                 Calories Carbohydrates     Fat Net Carbohydrates Saturated Fat
             id                   title readyInMinutes  .amount       .amount .amount           .amount       .amount
0       1089577  Hungarian Goulash Stew            120   323.18         38.95   15.14             34.64          4.43

对列进行排序，使标签在一起：

df = df.reindex(sorted(df.columns), axis=1)

.title                                                 Calories                            Carbohydrates
             id readyInMinutes                   title  .amount .percentOfDailyNeeds .unit       .amount .percentOfDailyNeeds .unit
0       1089577            120  Hungarian Goulash Stew   323.18                16.16  kcal         38.95                12.98     g

使用连接降低级别（创建 Calories.amount、Calories.unit 等）

df.columns = df.columns.map(''.join)

        id  readyInMinutes                   title  Calories.amount  Calories.percentOfDailyNeeds Calories.unit
0  1089577             120  Hungarian Goulash Stew           323.18                         16.16          kcal

Answer 2

您可以按如下方式使用df.pivot()：

(df.pivot(index=['id', 'title', 'readyInMinutes'], 
          columns='nutrients.title', 
          values='nutrients.amount')
          .rename_axis(None, axis=1)
).reset_index()

结果：

        id                   title  readyInMinutes  Calories  Carbohydrates    Fat  Net Carbohydrates  Saturated Fat
0  1089577  Hungarian Goulash Stew             120    323.18          38.95  15.14              34.64           4.43

在没有任何聚合的情况下在数据框中传播特定列？

Spread specific columns in dataframe without any aggregation?

python

python-3.x

pandas

reshape

melt