Pandas逐步过滤DataFrame：未定义过滤值的替代方案咨询

问题背景 #

我正在用Pandas搭建简易过滤系统，处理包含扑克手牌和玩家动作（call、raise、fold等）的数据库，想要实现一个通用过滤器，按玩家动作逐步过滤DataFrame：

• 过滤器包含action1、action2、action3、action4四个动作项
• 先将action1的过滤值设为"raise"（示例代码中用"c"作为测试值）
• action2、action3、action4暂未确定过滤条件
• 仅过滤action1，保留action2、action3、action4的所有可能值

但将action2、action3、action4设为空字符串时，过滤完全失效，请问应该用什么值替代？

附代码示例 #

import pandas as pd
import numpy as np

df = pd.read_csv('xxxxx', sep=";")

df.dropna(inplace = True)

###PREFLOP###

a1_preflop = "c"
a2_preflop = ""
a3_preflop = ""
a4_preflop = ""

x= df[(df.actions_1_preflop == a1_preflop) & (df.actions_2_preflop == a2_preflop) & (df.actions_3_preflop == a3_preflop)&(df.actions_4_preflop == a4_preflop)]

print(x)

未过滤的小型DataFrame示例（目标：过滤action1_preflop="c"，暂不过滤action2/3/4） #

myposition  tiers   besthand_flop   checker_flop    handtype_flop   topsuite_flop   topcolor_flop   besthand_turn   checker_turn    handtype_turn   topsuite_turn   topcolor_turn   besthand_river  checker_river   handtype_river  topsuite_river  topcolor_river  bet_1_preflop   bet_2_preflop   bet_3_preflop   bet_1_flop  bet_2_flop  bet_3_flop  bet_1_turn  bet_2_turn  bet_3_turn  bet_1_river bet_2_river bet_3_river action1_preflop action2_preflop action3_preflop action4_preflop action1_flop    action2_flop    action3_flop    action4_flop    action1_turn    action2_turn    action3_turn    action4_turn    action1_river   action2_river   action3_river   action4_river
bb  4   Brelan  1   high    3   3   Brelan  1   high    3   3   Brelan  1   high    3   3   1.3         4.6         193.1                       r   c           c   e   c       a   c           c   c       
sb  9   Double paire    0.5 very high   2   2   Double paire    0.5 very high   2   3   Double paire    0.5 very high   3   3   6                       14                      c   e   c       c   c           c   z   f                   
bb  9   Double paire    0.5 very high   2   3   Double paire    0.5 very high   3   3   Double paire    0.5 very high   4   3                           2           9   188     c   c           c   c           r   c           c   e   a   f
sb  4   Paire   1   high    4   2   Paire   1   high    4   2   Suite   0.8 very high   5   2   2                       1.5         5.85    189.65      c   r   c       c   c           r   c           e   a   f   
sb  9   Paire   1   high    2   2   Double paire    0.5 high    2   2   Double paire    0.5 high    3   3                           2                       c   c           c   c           r   c           c   c       
sb  9   Paire   1   high    3   2   Double paire    0.5 high    3   2   Brelan  0.666666667 high    3   3   3                       2           1.2         c   r   c       c   c           r   c           r   c       
bb  9   Paire   1   high    2   3   Double paire    0.5 high    2   3   Brelan  1   high    2   3               1.3                                 c   c           c   r   c       c   c           c   c       
bb  9   Paire   1   high    3   2   Paire   1   high    3   2   Double paire    0.5 very high   3   2   1.3                     197.7                       r   c           c   c           a   f                       
sb  5   Paire   1   high    3   2   Paire   1   high    3   2   Double paire    1   high    3   2                           1.3         197.7           c   c           c   c           r   c           c   a   f   
bb  7   Brelan  0.666666667 high    3   2   Brelan  0.666666667 high    3   3   Brelan  0.666666667 high    3   3   1.3                     4.6                     r   c           c   c           e   f                       

解决方案 #

核心逻辑是：当某个动作项不需要过滤时，让对应的条件返回全True，这样在逻辑与（&）运算中不会影响其他条件的筛选结果。以下是三种可行方案：

方案1：用布尔值True作为占位符 #

将不需要过滤的变量设为True，通过类型判断决定是否应用该条件：

import pandas as pd
import numpy as np

df = pd.read_csv('xxxxx', sep=";")
df.dropna(inplace=True)

###PREFLOP###
a1_preflop = "c"
a2_preflop = True  # 不需要过滤时设为True
a3_preflop = True
a4_preflop = True

# 构建过滤条件
cond1 = df.actions_1_preflop == a1_preflop
cond2 = df.actions_2_preflop == a2_preflop if not isinstance(a2_preflop, bool) else True
cond3 = df.actions_3_preflop == a3_preflop if not isinstance(a3_preflop, bool) else True
cond4 = df.actions_4_preflop == a4_preflop if not isinstance(a4_preflop, bool) else True

x = df[cond1 & cond2 & cond3 & cond4]
print(x)

方案2：动态构建过滤条件列表 #

只添加需要过滤的条件，避免冗余判断，扩展性最强：

import pandas as pd
import numpy as np

df = pd.read_csv('xxxxx', sep=";")
df.dropna(inplace=True)

###PREFLOP###
filters = []
# 添加需要过滤的条件
filters.append(df.actions_1_preflop == "c")
# 后续需要过滤action2时，直接追加条件即可
# filters.append(df.actions_2_preflop == "raise")
# filters.append(df.actions_3_preflop == "fold")

# 合并所有条件
x = df[np.all(filters, axis=0)]
print(x)

方案3：用None作为占位符 #

将不需要过滤的变量设为None，通过判断变量是否为None来决定是否应用条件：

import pandas as pd
import numpy as np

df = pd.read_csv('xxxxx', sep=";")
df.dropna(inplace=True)

###PREFLOP###
a1_preflop = "c"
a2_preflop = None
a3_preflop = None
a4_preflop = None

cond1 = df.actions_1_preflop == a1_preflop
cond2 = df.actions_2_preflop == a2_preflop if a2_preflop is not None else True
cond3 = df.actions_3_preflop == a3_preflop if a3_preflop is not None else True
cond4 = df.actions_4_preflop == a4_preflop if a4_preflop is not None else True

x = df[cond1 & cond2 & cond3 & cond4]
print(x)

以上三种方案都能实现仅过滤指定动作项、保留其他动作所有可能值的需求，其中方案2的灵活性最高，后续添加新过滤条件只需追加列表项即可。

内容的提问来源于Stack Exchange，提问作者Raphaël Ambit