Question

在下面的给定数据框中，我要清理计数器（例如MEM_TRANS_RETIRED），并将列load,rps,th95

添加到数据框中

       load        rps     th95    energy     ...       MEM_TRANS_RETIRED-34  PERF_COUNT_HW_CACHE_L1D-34  PERF_COUNT_HW_CACHE_L1I-34     map_freq
0    500.0k  346222.62    12.62   7270.22     ...                  154287.14                   591053.74                2.918521e+07   6C-1.70GHz
1    400.0k  402628.34     2.25  12026.40     ...                  189915.07                   627043.91                2.867945e+07  10C-2.10GHz
2    500.0k  283508.27    15.52   5662.74     ...                  140790.31                  1431892.98                4.253950e+07   6C-1.30GHz

这就是我的做法

self.unique_counters = [x[:-2] for x in self.dfile_keys[6:] if x.endswith('-0')]
for counter in self.unique_counters:
 new = self.dfile.loc[:, self.dfile.columns.str.startswith(counter)]

但是，这仅提供选定的列，而没有上面提到的其他列。

     PERF_COUNT_HW_CPU_CYCLES-0  PERF_COUNT_HW_CPU_CYCLES-2             ...               PERF_COUNT_HW_CPU_CYCLES-32  PERF_COUNT_HW_CPU_CYCLES-34
0                  6.020913e+08                6.021277e+08             ...                              5.109342e+06                 2.556039e+06
1                  4.781879e+08                4.783621e+08             ...                              3.095814e+06                 2.795868e+06
2                  4.841784e+08                4.844846e+08             ...                              2.389396e+06                 5.550159e+06

如何同时获得带有选定起始字符串和一些指定列的计数器这是预期的输出

       load        rps     th95    energy   PERF_COUNT_HW_CPU_CYCLES-0  PERF_COUNT_HW_CPU_CYCLES-2             ...               PERF_COUNT_HW_CPU_CYCLES-32  PERF_COUNT_HW_CPU_CYCLES-34
0    500.0k  346222.62    12.62   7270.22   6.020913e+08                6.021277e+08             ...                              5.109342e+06                 2.556039e+06
1    400.0k  402628.34     2.25  12026.40   4.781879e+08                4.783621e+08             ...                              3.095814e+06                 2.795868e+06
2    500.0k  283508.27    15.52   5662.74   4.841784e+08                4.844846e+08             ...                              2.389396e+06                 5.550159e+06

Answer 1

我相信您需要一个新的DataFrame：

L = [x[:-2] for x in self.dfile_keys[6:] if x.endswith('-0')]

new = self.dfile.loc[:, self.dfile.columns.str.startswith(tuple(L))]

df = pd.concat([self.dfile[['load','rps','th95','energy']], new], axis=1)

或者在列表理解中创建list of DataFrame：

self.unique_counters = [x[:-2] for x in self.dfile_keys[6:] if x.endswith('-0')]
dfs = [self.dfile.loc[:, self.dfile.columns.str.startswith(counter)] 
       for counter in self.unique_counters]

df = pd.concat([self.dfile[['load','rps','th95','energy']], dfs], axis=1)

pandas数据框-具有选定起始字符串和指定列的列

1 个答案: