python - 根据列值过滤多索引数据框，删除级别内的所有行

Question

我正在尝试根据一个或多个值过滤 DataFrame。这是一个 CSV 示例:

AlignmentId,TranscriptId,classifier,value
ENSMUST00000025010-1,ENSMUST00000025010,AlnCoverage,0.99612
ENSMUST00000025010-1,ENSMUST00000025010,AlnIdentity,0.93553
ENSMUST00000025010-1,ENSMUST00000025010,Badness,0.06749
ENSMUST00000025014-1,ENSMUST00000025014,AlnCoverage,1.0
ENSMUST00000025014-1,ENSMUST00000025014,AlnIdentity,0.96382
ENSMUST00000025014-1,ENSMUST00000025014,Badness,0.03618

加载时:

>>> df = pd.read_csv('tmp.csv', index_col=['AlignmentId', 'TranscriptId'])
>>> df
                                          classifier    value
AlignmentId          TranscriptId
ENSMUST00000025010-1 ENSMUST00000025010  AlnCoverage  0.99612
                     ENSMUST00000025010  AlnIdentity  0.93553
                     ENSMUST00000025010      Badness  0.06749
ENSMUST00000025014-1 ENSMUST00000025014  AlnCoverage  1.00000
                     ENSMUST00000025014  AlnIdentity  0.96382
                     ENSMUST00000025014      Badness  0.03618

我想删除每个 AlignmentId未能通过一系列 classifiers 的组.对于这个例子，假设我想删除 ENSMUST00000025010因为AlnCoverage < 1.0 .因此，我想以这个数据框结束:

ENSMUST00000025014-1 ENSMUST00000025014  AlnCoverage  1.00000
                     ENSMUST00000025014  AlnIdentity  0.96382
                     ENSMUST00000025014      Badness  0.03618

我该怎么做？

Answer 1

试试这个:

In [169]: df = df.drop(df[(df.classifier=='AlnCoverage') & (df.value < 1)].index)

In [170]: df
Out[170]:
                                          classifier    value
AlignmentId          TranscriptId
ENSMUST00000025014-1 ENSMUST00000025014  AlnCoverage  1.00000
                     ENSMUST00000025014  AlnIdentity  0.96382
                     ENSMUST00000025014      Badness  0.03618