pandas - 从字典和索引顺序创建 DataFrame？

Question

我正在使用最近下载的 Anaconda 和 Python 3.7.1 & pandas 0.23.4

pandas doc说:

When the data is a dict, and an index is not passed, the Series index will be ordered by the dict’s insertion order

我从没有传递索引的字典中实例化一个 pandas DataFrame:

newspapers = {'Jim':{'Mon':15,'Tue':17,'Wed':21,'Thu':16,'Fri':19},\
     'Tony':{'Mon':8,'Tue':15,'Wed':11,'Thu':16,'Fri':13}, \
     'Colin':{'Mon':13,'Tue':17,'Wed':19,'Thu':17,'Fri':20} \
    }
newspapers_df = pd.DataFrame(newspapers)

为什么这不按插入顺序显示，星期一、星期二、星期三、星期四、星期五？:

print(newspapers_df)

输出:

      Jim  Tony  Colin
Fri   19    13     20
Mon   15     8     13
Thu   16    16     17
Tue   17    15     17
Wed   21    11     19

Answer 1

这似乎是错误，对于我在 python 3.5 中工作，pandas 0.24.2 在字典理解中创建 Series 并传递给 DataFrame 构造函数:

newspapers_df = pd.DataFrame({k:pd.Series(v) for k, v in newspapers.items()})

print (newspapers_df)
     Jim  Tony  Colin
Mon   15     8     13
Tue   17    15     17
Wed   21    11     19
Thu   16    16     17
Fri   19    13     20

---

您的数据的可能解决方案 - DataFrame.reindex或 排序的 CategoricalIndex:

newspapers_df = pd.DataFrame(newspapers)

L = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri']
newspapers_df = newspapers_df.reindex(L)

或者:

newspapers_df.index = pd.CategoricalIndex(newspapers_df.index, ordered=True, categories=L)
newspapers_df = newspapers_df.sort_index()
print (newspapers_df)
     Jim  Tony  Colin
Mon   15     8     13
Tue   17    15     17
Wed   21    11     19
Thu   16    16     17
Fri   19    13     20