pandas-向数据框添加一系列会导致出现NaN值

Question

我有一个看起来像这样的数据框:

d = {'Col_1' : pd.Series(['A', 'A', 'A', 'B']),
     'Col_2' : pd.Series(['B', 'C', 'B', 'D']),
     'Col_3' : pd.Series([np.nan, 'D', 'C', np.nan]),
     'Col_4' : pd.Series([np.nan, np.nan, 'D', np.nan]),
     'Col_5' : pd.Series([np.nan, np.nan, 'E', np.nan]),}
df = pd.DataFrame(d)

Col_1  Col_2  Col_3  Col_4  Col_5
  A      B      NaN    NaN    NaN
  A      C      D      NaN    NaN
  A      B      C      D      E
  B      D      NaN    NaN    NaN

我的目标是最终得到以下内容:

Col_1  Col_2  Col_3  Col_4  Col_5  ConCat
  A      B      NaN    NaN    NaN    A:B
  A      C      D      NaN    NaN    A:C:D
  A      B      C      D      E      A:B:C:D:E
  B      D      NaN    NaN    NaN    B:D

我已经成功创建了一个看起来像所需输出的数据框:

rows = df.values
df_1 = pd.DataFrame([':'.join(word for word in rows if word is not np.nan) for rows in rows])

    0
0  A:B
1  A:C:D
2  A:B:C:D:E
3  B:D

但是现在当我尝试将它放入原始数据帧时,我得到:

df['concatenated'] = df_1

Col_1  Col_2  Col_3  Col_4  Col_5  concatenated
  A      B      NaN    NaN    NaN    NaN
  A      C      D      NaN    NaN    NaN
  A      B      C      D      E      NaN
  B      D      NaN    NaN    NaN    NaN

奇怪的是,在创建简化示例时,它按预期工作.如果我正在做的完整代码如下.原始数据来自我上面原始数据框的转换.

df_caregiver_type = pd.concat([df_caregiver_type[col].order().reset_index(drop=True) for col in df_caregiver_type], axis=1, ignore_index=False).T
df_caregiver_type.rename(columns=lambda x: 'Col_' + str(x), inplace=True)
rows = df_caregiver_type.values
df_caregiver_type1 = pd.DataFrame([':'.join(word for word in rows if word is not np.nan) for rows in rows])
df_caregiver_type['concatenated'] = df_caregiver_type1
df_caregiver_type = df_caregiver_type.T
df_caregiver_type

更新 我认为由于完整代码的第一行,我收到错误.它来自一个单独但相关的问题:pandas:单独对每列进行排序

Answer 1

对于您的完整数据集,将最后一步更改df['concatenated'] = df_1为df['concatenated'] = df_1.values将解决问题,我认为这是一个错误,我非常确定我之前已经看过它.

要不就: df['concatenated'] = [':'.join(word for word in row if word is not np.nan) for row in rows]