Question

我有以下数据框：

df:
      y     m   d   val
0   2013    10  1   33.5
1   2013    10  2   37.1
2   2013    10  3   25.9
3   2013    10  4   31.3
4   2013    10  5   35.3
5   2013    10  6   55.4
6   2013    10  7   29.5
7   2013    10  8   31.3
8   2013    10  9   27.7
9   2013    10  10  25.9

其中y，m，d分别对应年份，月份和日期。我想汇总它们并转换为datetime。

df['date'] =  0
for v in df.index:
    df['date'][v] = datetime.datetime(df.y[v], df.m[v], df.d[v])

我想知道哪种方法可以避免这种循环

Answer 1

来自docstring：

Assembling a datetime from multiple columns of a DataFrame. The keys can be
common abbreviations like ['year', 'month', 'day', 'minute', 'second',
'ms', 'us', 'ns']) or plurals of the same

>>> df = pd.DataFrame({'year': [2015, 2016],
                       'month': [2, 3],
                       'day': [4, 5]})
>>> pd.to_datetime(df)
0   2015-02-04
1   2016-03-05
dtype: datetime64[ns]

代码：

In [135]: pd.to_datetime(df.rename(columns={'y':'Year','m':'Month','d':'Day'}).iloc[:, :3])
Out[135]:
0   2013-10-01
1   2013-10-02
2   2013-10-03
3   2013-10-04
4   2013-10-05
5   2013-10-06
6   2013-10-07
7   2013-10-08
8   2013-10-09
9   2013-10-10
dtype: datetime64[ns]

Answer 2

这是一种方式：

pd.to_datetime((df['y']*10000 + df['m']*100 + df['d']).astype(str))
Out: 
0   2013-10-01
1   2013-10-02
2   2013-10-03
3   2013-10-04
4   2013-10-05
5   2013-10-06
6   2013-10-07
7   2013-10-08
8   2013-10-09
9   2013-10-10
dtype: datetime64[ns]

Python：如何避免在pandas数据帧中转换循环？

2 个答案: