我想转换foll。数据：
mapStateToProps到长度为jan_1 jan_15 feb_1 feb_15 mar_1 mar_15 apr_1 apr_15 may_1 may_15 jun_1 jun_15 jul_1 jul_15 aug_1 aug_15 sep_1 sep_15 oct_1 oct_15 nov_1 nov_15 dec_1 dec_15
0 0 0 0 0 1 1 2 2 2 2 2 2 3 3 3 3 3 0 0 0 0 0 0
的数组中，其中每个元素重复到下一个日期，例如从365到january 1 ...
我可以执行类似january 15的操作，但这不是日期识别，因此不会考虑numpy.repeat和feb_15之间发生的时间少于15天。
任何pythonic解决方案？
答案 0 :(得分：2)
IIUC你可以这样做：
In [194]: %paste
# transpose DF, rename columns
x = df.T.reset_index().rename(columns={'index':'date', 0:'val'})
# parse dates
x['date'] = pd.to_datetime(x['date'], format='%b_%d')
# group resampled DF by the month and resample(`D`) each group
result = (x.groupby(x['date'].dt.month)
.apply(lambda x: x.set_index('date').resample('1D').ffill()))
# rename index names
result.index.names = ['month','date']
## -- End pasted text --
In [212]: result
Out[212]:
val
month date
1 1900-01-01 0
1900-01-02 0
1900-01-03 0
1900-01-04 0
1900-01-05 0
1900-01-06 0
1900-01-07 0
1900-01-08 0
1900-01-09 0
1900-01-10 0
1900-01-11 0
1900-01-12 0
1900-01-13 0
1900-01-14 0
1900-01-15 0
2 1900-02-01 0
1900-02-02 0
1900-02-03 0
1900-02-04 0
1900-02-05 0
1900-02-06 0
1900-02-07 0
1900-02-08 0
1900-02-09 0
1900-02-10 0
1900-02-11 0
1900-02-12 0
1900-02-13 0
1900-02-14 0
1900-02-15 0
... ...
11 1900-11-01 0
1900-11-02 0
1900-11-03 0
1900-11-04 0
1900-11-05 0
1900-11-06 0
1900-11-07 0
1900-11-08 0
1900-11-09 0
1900-11-10 0
1900-11-11 0
1900-11-12 0
1900-11-13 0
1900-11-14 0
1900-11-15 0
12 1900-12-01 0
1900-12-02 0
1900-12-03 0
1900-12-04 0
1900-12-05 0
1900-12-06 0
1900-12-07 0
1900-12-08 0
1900-12-09 0
1900-12-10 0
1900-12-11 0
1900-12-12 0
1900-12-13 0
1900-12-14 0
1900-12-15 0
[180 rows x 1 columns]
或使用reset_index()：
In [213]: result.reset_index().head(20)
Out[213]:
month date val
0 1 1900-01-01 0
1 1 1900-01-02 0
2 1 1900-01-03 0
3 1 1900-01-04 0
4 1 1900-01-05 0
5 1 1900-01-06 0
6 1 1900-01-07 0
7 1 1900-01-08 0
8 1 1900-01-09 0
9 1 1900-01-10 0
10 1 1900-01-11 0
11 1 1900-01-12 0
12 1 1900-01-13 0
13 1 1900-01-14 0
14 1 1900-01-15 0
15 2 1900-02-01 0
16 2 1900-02-02 0
17 2 1900-02-03 0
18 2 1900-02-04 0
19 2 1900-02-05 0
答案 1 :(得分：1)
您可以使用resample：
#add last value - 31 dec by value of last column of df
df['dec_31'] = df.iloc[:,-1]
#convert to datetime - see http://strftime.org/
df.columns = pd.to_datetime(df.columns, format='%b_%d')
#transpose and resample by days
df1 = df.T.resample('d').ffill()
df1.columns = ['col']
print (df1)
col
1900-01-01 0
1900-01-02 0
1900-01-03 0
1900-01-04 0
1900-01-05 0
1900-01-06 0
1900-01-07 0
1900-01-08 0
1900-01-09 0
1900-01-10 0
1900-01-11 0
1900-01-12 0
1900-01-13 0
1900-01-14 0
1900-01-15 0
1900-01-16 0
1900-01-17 0
1900-01-18 0
1900-01-19 0
1900-01-20 0
1900-01-21 0
1900-01-22 0
1900-01-23 0
1900-01-24 0
1900-01-25 0
1900-01-26 0
1900-01-27 0
1900-01-28 0
1900-01-29 0
1900-01-30 0
..
1900-12-02 0
1900-12-03 0
1900-12-04 0
1900-12-05 0
1900-12-06 0
1900-12-07 0
1900-12-08 0
1900-12-09 0
1900-12-10 0
1900-12-11 0
1900-12-12 0
1900-12-13 0
1900-12-14 0
1900-12-15 0
1900-12-16 0
1900-12-17 0
1900-12-18 0
1900-12-19 0
1900-12-20 0
1900-12-21 0
1900-12-22 0
1900-12-23 0
1900-12-24 0
1900-12-25 0
1900-12-26 0
1900-12-27 0
1900-12-28 0
1900-12-29 0
1900-12-30 0
1900-12-31 0
[365 rows x 1 columns]
#if need serie
print (df1.col)
1900-01-01 0
1900-01-02 0
1900-01-03 0
1900-01-04 0
1900-01-05 0
1900-01-06 0
1900-01-07 0
1900-01-08 0
1900-01-09 0
1900-01-10 0
1900-01-11 0
1900-01-12 0
1900-01-13 0
1900-01-14 0
1900-01-15 0
1900-01-16 0
1900-01-17 0
1900-01-18 0
1900-01-19 0
1900-01-20 0
1900-01-21 0
1900-01-22 0
1900-01-23 0
1900-01-24 0
1900-01-25 0
1900-01-26 0
1900-01-27 0
1900-01-28 0
1900-01-29 0
1900-01-30 0
..
1900-12-02 0
1900-12-03 0
1900-12-04 0
1900-12-05 0
1900-12-06 0
1900-12-07 0
1900-12-08 0
1900-12-09 0
1900-12-10 0
1900-12-11 0
1900-12-12 0
1900-12-13 0
1900-12-14 0
1900-12-15 0
1900-12-16 0
1900-12-17 0
1900-12-18 0
1900-12-19 0
1900-12-20 0
1900-12-21 0
1900-12-22 0
1900-12-23 0
1900-12-24 0
1900-12-25 0
1900-12-26 0
1900-12-27 0
1900-12-28 0
1900-12-29 0
1900-12-30 0
1900-12-31 0
Freq: D, Name: col, dtype: int64
#transpose and convert to numpy array
print (df1.T.values)
[[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3 3 3 3 3 3 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]]