Python For循环内嵌套IF语句为Pandas DataFrame新增列时elif不执行

问题根源 #

你代码的核心问题出在赋值逻辑上：

• 遍历Agent列的每个值时，每次赋值都是直接修改整个AgentTag列，而非修改当前遍历对应的行。假设第一个遍历到的Agent值属于第一组，整个列会被直接全部设为Agent 1，就算后续遍历到其他组的Agent值，后续的赋值也会覆盖全列，最终全列的值由最后一个遍历到的Agent值所属分组决定，看起来就像只走了第一个分支。
• 其次这种逐行遍历的写法在pandas中效率极低，数据量稍大就会非常慢，完全没必要用循环实现。

正确实现方案 #

方案1：字典映射（最简洁，推荐） #

先构造Agent到AgentTag的映射字典，直接用map方法生成新列：

# 构造映射关系
agent_map = {
    "unez": "Agent 1",
    "rmbua": "Agent 1",
    "destrada": "Agent 1",
    "amateo": "Agent 1",
    "cmabelison": "Agent 1",
    "rverga": "Agent 2",
    "dpcaban": "Agent 2",
    "dgsugui": "Agent 2",
    "gmic": "Agent 3",
    "jdera": "Agent 3",
    "gras": "Agent 4",
    "mcsrra": "Agent 4",
    "jcawan": "Agent 5",
    "rmcola": "Agent 5",
    "mjgamo": "Agent 5",
    "ychaco": "Agent 6",
    "phondra": "Agent 6",
    "mmorang": "Agent 7",
    "vsin": "Agent 7",
    "pbong": "Agent 8"
}
# 生成新列
df_csrdata_2mos_Filtered_Done["AgentTag"] = df_csrdata_2mos_Filtered_Done["Agent"].map(agent_map)
print("AgentTag Done!")

未匹配到的Agent对应行的AgentTag会自动设为NaN，可后续按需填充默认值。

方案2：np.select（适合条件更复杂的场景） #

如果后续分组逻辑需要调整得更复杂，可以用numpy的select方法实现：

import numpy as np

# 定义条件列表
conditions = [
    df_csrdata_2mos_Filtered_Done["Agent"].isin(["unez", "rmbua", "destrada", "amateo", "cmabelison"]),
    df_csrdata_2mos_Filtered_Done["Agent"].isin(["rverga", "dpcaban", "dgsugui"]),
    df_csrdata_2mos_Filtered_Done["Agent"].isin(["gmic", "jdera"]),
    df_csrdata_2mos_Filtered_Done["Agent"].isin(["gras", "mcsrra"]),
    df_csrdata_2mos_Filtered_Done["Agent"].isin(["jcawan", "rmcola", "mjgamo"]),
    df_csrdata_2mos_Filtered_Done["Agent"].isin(["ychaco", "phondra"]),
    df_csrdata_2mos_Filtered_Done["Agent"].isin(["mmorang", "vsin"]),
    df_csrdata_2mos_Filtered_Done["Agent"] == "pbong"
]
# 定义对应结果
choices = ["Agent 1", "Agent 2", "Agent 3", "Agent 4", "Agent 5", "Agent 6", "Agent 7", "Agent 8"]
# 生成新列，可修改default参数设置未匹配项的默认值
df_csrdata_2mos_Filtered_Done["AgentTag"] = np.select(conditions, choices, default=np.nan)
print("AgentTag Done!")

内容的提问来源于stack exchange，提问作者COCO