使用Python3中的条件从Pandas获取行

Question

我有一个包含以下列的数据格式。

CUI         id          term id term_name 
C0000729    10000057    MDR LLT Abdominal cramps
C0000729    10000056    MDR LLT Abdominal cramp
C0000729    10011286    MDR LLT Cramp abdominal
C0000729    10000058    MDR LLT Abdominal crampy pains
C0000729    10093764    ICD10   PT  Abdominal crampy pains
C0000800    10093765    ICD10   PT  Abdominal pain
C0000800    10000058    MDR LLT Abdominal crampy pains
C0000800    10093764    ICD10AM PT  Abdominal crampy pains
C0000730    10000052    MDR LLT Abdominal cramps back

如果术语是'MDR‘和和ICDs (ICD10，ICD10AM)，我想获取这些行(每个CUI)。但是，如果它只是MDR，排除它(例如。最后一行'C0000730')。

预期产出如下：

CUI         id          term id term_name 
C0000729    10000057    MDR LLT Abdominal cramps
C0000729    10000056    MDR LLT Abdominal cramp
C0000729    10011286    MDR LLT Cramp abdominal
C0000729    10000058    MDR LLT Abdominal crampy pains
C0000729    10093764    ICD10   PT  Abdominal crampy pains
C0000800    10093765    ICD10   PT  Abdominal pain
C0000800    10000058    MDR LLT Abdominal crampy pains
C0000800    10093764    ICD10AM PT  Abdominal crampy pains

我正在使用下面的代码行使用上面的数据。

#select only those mappings where ICD and MeDRA both exists for a particular CUI id
s = set(['ICD10','ICD10CM','ICD10AM','MDR'])

dff_mapped = df_umls[df_umls.groupby('CUI')['SAB'].transform(lambda x: set(x) == s)]

dff_mapped = dff_mapped.sort_values(['CUI', 'SAB'],ascending = [True, True])

dff_mapped.to_csv('df_mapped', index = False, sep = ',')

任何帮助都是非常感谢的。

Answer 1

如果我理解得对：

nonMDR = df[df['term id'].str.startswith('ICD')] # creates a new df with ICDs
term_ids = nonMDR['CUI'].unique() # create an array of unique CUIs
df[df['CUI'].isin(term_ids)] # filter CUIs