Question

我有一个像这样的数据框：

band    mean    raster
1   894.343482  D:/Python/Copied/selection/20170219_095504.tif
2   1159.282304 D:/Python/Copied/selection/20170219_095504.tif
3   1342.291595 D:/Python/Copied/selection/20170219_095504.tif
4   3056.809463 D:/Python/Copied/selection/20170219_095504.tif
1   516.9624071 D:/Python/Copied/selection/20170325_095551.tif
2   720.1932533 D:/Python/Copied/selection/20170325_095551.tif
3   689.6287879 D:/Python/Copied/selection/20170325_095551.tif
4   4561.576329 D:/Python/Copied/selection/20170325_095551.tif
1   566.2016867 D:/Python/Copied/selection/20170527_095700.tif
2   812.9927101 D:/Python/Copied/selection/20170527_095700.tif
3   760.4621212 D:/Python/Copied/selection/20170527_095700.tif
4   5009.537164 D:/Python/Copied/selection/20170527_095700.tif

我想将其格式化为：

band1_mean  band2_mean  band3_mean  band4_mean  raster_name         id
894.343482  1159.282304 1342.291595 3056.809463 20170219_095504.tif 1
516.9624071 720.1932533 689.6287879 4561.576329 20170325_095551.tif 2
566.2016867 812.9927101 760.4621212 5009.537164 20170527_095700.tif 3

所有4个波段都属于一个栅格，因此值必须全部放在一行中。我不知道如何在没有每个栅格的键ID的情况下堆叠它们。谢谢！

Answer 1

有了df.pivot("raster", "band", "mean")，您将得到

band                          1            2            3            4
raster                                                                
20170219_095504.tif  894.343482  1159.282304  1342.291595  3056.809463
20170325_095551.tif  516.962407   720.193253   689.628788  4561.576329
20170527_095700.tif  566.201687   812.992710   760.462121  5009.537164

Answer 2

这是pivot的情况：

# extract the raster name:
df['raster_name'] = df.raster.str.extract('(\d+_\d+\.tif)')

# pivot
new_df = df.pivot(index='raster_name', columns='band', values='mean')

# rename the columns:
new_df.columns = [f'band{i}_mean' for i in new_df.columns]

输出：

                     band1_mean   band2_mean   band3_mean   band4_mean
raster_name                                                           
20170219_095504.tif  894.343482  1159.282304  1342.291595  3056.809463
20170325_095551.tif  516.962407   720.193253   689.628788  4561.576329
20170527_095700.tif  566.201687   812.992710   760.462121  5009.537164

如果您希望reset_index是普通列，则可以在new_df上raster_name。

如何将行数堆叠为一行并分配ID

2 个答案: