我有一个pandas数据框,其中包含单词嵌入。单词是列名。为了获得最终的句子嵌入,我这样做:
st = ['some','random','text']
a = df[st].sum(axis=1)
当所有术语都出现在数据框中时,这很好用。但是,当数据帧中不存在该列时,它将引发密钥错误。目前,我正在为每个学期循环学习,除了尝试。我想避免循环,并在一行中执行该操作。熊猫提供相同的选项吗?
答案 0 :(得分:1)
使用intersection
:
df[df.columns.intersection(st)].sum(axis=1)
这只会对列在df
中的列求和
答案 1 :(得分:0)
您可以将任务替换为:
@InjectMocks
@Spy
private ContractServiceImpl service;
@Mock
ContractRepository contractRepository;
@Mock
OasysContractRepository oasysContractRepository;
@Mock
ContractServiceManager serviceManager;
public static final String contractNumber = "3900006835";
@Test
public void getListOfContractBy_hcp_success() {
List<OasysContract> oasysContractList = new ArrayList<>();
oasysContractList.isEmpty();
List<Contract> contractList = new ArrayList<>();
contractList.add(BuildUtil.buildContract());
//mock
Mockito.doReturn(oasysContractList).when(oasysContractRepository).findByContractNumber(contractNumber);
Mockito.doReturn(BuildUtil.buildContractList()).when(contractRepository).findContractsByContractNumber(contractNumber);
Mockito.doReturn(contractList).when(serviceManager).checkPgwContract(contractList);
Mockito.doReturn(BuildUtil.buildContract()).when(contractRepository).findContractByExpiredDate(contractNumber);
Mockito.doReturn("N").when(contractRepository).getFullyPaid(contractNumber);
//test
List<Contract> contracts = service.getListOfContractBy(contractNumber);
System.out.println(contracts);
assert(!contracts.isEmpty());
}
消除所有不存在的单词。