多进程循环(python)

我有一个数据帧,我正在尝试使用多进程来更有效地使用可用的核心进行循环。有一个例子 here 然而,我不知道“pool”是如何应用的。

示例df代码:

df = pd.DataFrame({
    'mac':['type_a','type_a','type_a','type_a','type_a','type_b','type_b','type_b','type_b','type_b'],
    'con':['a','a','a','c','b','a','a','b','a','c'],
    'result':[1,1,2,2,3,1,1,3,1,2],
})

MAX_NUMBER = 3 
for j in range(MAX_NUMBER):
    
    i = j + 1 
    aft = f"mc{i}_aft"
    bef = f"mc{i}_bef"
    
    holder1 = ['con', 'mac', 'result']
    holder2 = ['con', 'mac']
    df['add'] = (df.groupby(holder1).cumcount() + 1)
    df[aft] = df['add'].loc[df.result == i]

    df = df.groupby(holder2, group_keys=False).apply(lambda x: x.fillna(method='ffill').fillna(0))

    df = df.drop(['add'], axis=1)
    df.fillna(0, inplace=True)
    df[aft] = df[aft].astype('int')
    
    df[bef] = df.groupby(holder2)[aft].shift(1)
    df.fillna(0, inplace=True)
    df[bef] = df[bef].astype('int')

示例df:

╔═══╦════════╦═════╦════════╗
║   ║  mac   ║ con ║ result ║
╠═══╬════════╬═════╬════════╣
║ 0 ║ type_a ║ a   ║      1 ║
║ 1 ║ type_a ║ a   ║      1 ║
║ 2 ║ type_a ║ a   ║      2 ║
║ 3 ║ type_a ║ c   ║      2 ║
║ 4 ║ type_a ║ b   ║      3 ║
║ 5 ║ type_b ║ a   ║      1 ║
║ 6 ║ type_b ║ a   ║      1 ║
║ 7 ║ type_b ║ b   ║      3 ║
║ 8 ║ type_b ║ a   ║      1 ║
║ 9 ║ type_b ║ c   ║      2 ║
╚═══╩════════╩═════╩════════╝

样本输出:

╔═══╦════════╦═════╦════════╦═════════╦═════════╦═════════╦═════════╦═════════╦═════════╗
║   ║  mac   ║ con ║ result ║ mc1_aft ║ mc1_bef ║ mc2_aft ║ mc2_bef ║ mc3_aft ║ mc3_bef ║
╠═══╬════════╬═════╬════════╬═════════╬═════════╬═════════╬═════════╬═════════╬═════════╣
║ 0 ║ type_a ║ a   ║      1 ║       1 ║       0 ║       0 ║       0 ║       0 ║       0 ║
║ 1 ║ type_a ║ a   ║      1 ║       2 ║       1 ║       0 ║       0 ║       0 ║       0 ║
║ 2 ║ type_a ║ a   ║      2 ║       2 ║       2 ║       1 ║       0 ║       0 ║       0 ║
║ 3 ║ type_a ║ c   ║      2 ║       0 ║       0 ║       1 ║       0 ║       0 ║       0 ║
║ 4 ║ type_a ║ b   ║      3 ║       0 ║       0 ║       0 ║       0 ║       1 ║       0 ║
║ 5 ║ type_b ║ a   ║      1 ║       1 ║       0 ║       0 ║       0 ║       0 ║       0 ║
║ 6 ║ type_b ║ a   ║      1 ║       2 ║       1 ║       0 ║       0 ║       0 ║       0 ║
║ 7 ║ type_b ║ b   ║      3 ║       0 ║       0 ║       0 ║       0 ║       1 ║       0 ║
║ 8 ║ type_b ║ a   ║      1 ║       3 ║       2 ║       0 ║       0 ║       0 ║       0 ║
║ 9 ║ type_b ║ c   ║      2 ║       0 ║       0 ║       1 ║       0 ║       0 ║       0 ║
╚═══╩════════╩═════╩════════╩═════════╩═════════╩═════════╩═════════╩═════════╩═════════╝

ps.ectual df要大得多,涉及更多的变量和循环,因此,尝试充分利用可用的内核来节省处理时间。循环按测试前(bef)和测试后(aft)的结果/累计数分组。

恳请