Question

我是Python编码的新手。目前，我正在尝试分析包含多个工作流程的数据框。每个工作流程都有用于启动和结束工作流程的不同处理步骤。在简化版本中，我的数据如下所示：

   Workflow Initiate   End_1   End_2   End_3
0         1   Name_1      na  Name_1      na
1         2   Name_2      na      na      na
2         3   Name_3      na      na  Name_5
3         4   Name_4  Name_5      na      na
4         5       na      na      na  Name_5

对于每个工作流程，我想比较结束工作流程的名称和启动工作流程的名称是否不同。

通过以下方式遍历各行，可以得到所需的输出：

for index, row in df.iterrows():
    if ((row['Initiate'] != 'na')
        and (row['Initiate'] == row['End_1']) |
            (row['Initiate'] == row['End_2']) |
            (row['Initiate'] == row['End_3'])
        ):
        print("Name end equals initiate")
    elif ((row['End_1'] == 'na') &
          (row['End_2'] == 'na') &
          (row['End_3'] == 'na')
         ):
        print("No name ended")
    else:
        print("Different name ended")

Name end equals initiate
No name ended
Different name ended
Different name ended
Different name ended

但是，我想在数据框中添加一列“分析”，以显示每个工作流程背后的上述结果。

为此，我将代码填充到一个函数中：

def function_name(a, b, c, d):
    for index, row in df.iterrows():
        if ((a != 'na')
            and (a == b) |
                (a == c) |
                (a == d)
            ):
            return "Name end equals initiate"
        elif ((b == 'na') &
              (c == 'na') &
              (d == 'na')
             ):
            return "No name ended"
        else:
            return "Different name ended"

df['Analysis'] = function_name(row['Initiate'],
                               row['End_1'],
                               row['End_2'],
                               row['End_3'])

print(df)

   Workflow Initiate          ...            End_3              Analysis
0         1   Name_1          ...               na  Different name ended
1         2   Name_2          ...               na  Different name ended
2         3   Name_3          ...           Name_5  Different name ended
3         4   Name_4          ...               na  Different name ended
4         5       na          ...           Name_5  Different name ended

您可以看到输出与第一个分析不同。我想在数据框中添加一个额外的列，该列为我提供与print语句相同的输出。

Answer 1

您应该在此处避免按行循环。您的算法是矢量化的：

df = df.replace('na', np.nan)  # replace string 'na' with NaN for efficient processing
ends = df.filter(like='End')  # filter by columns with 'End'

match = ends.ffill(1).iloc[:, -1] == df['Initiate']  # find last Name in each End
nulls = ends.isnull().all(1)  # check which rows are all null

# apply vectorised conditional logic
df['Result'] = np.select([match, nulls], ['Name end equals initiate', 'No name ended'],
                         'Different name ended')

print(df)

   Workflow Initiate   End_1   End_2   End_3                    Result
0         1   Name_1     NaN  Name_1     NaN  Name end equals initiate
1         2   Name_2     NaN     NaN     NaN             No name ended
2         3   Name_3     NaN     NaN  Name_5      Different name ended
3         4   Name_4  Name_5     NaN     NaN      Different name ended
4         5      NaN     NaN     NaN  Name_5      Different name ended

Answer 2

尝试使用np.select（）

conditions = [
    (df['Initiate'] != 'na') & ((df['Initiate'] == df['End_1']) | (df['Initiate'] == df['End_2']) | (df['Initiate'] == df['End_3'])),
    (df['End_1'] == 'na') & (df['End_2'] == 'na') & (df['End_3'] == 'na')
] 

answers = ['Name end equals initiate','No name ended']   
df['Analysis'] = np.select(conditions, answers, default='Different name ended')

如何在新列中的Pandas DataFrame行上存储迭代结果？

2 个答案: