使用 Pandas 过滤基于列的特定信息

解释：

1. 创建仅包含列“y”中的“特殊”和“其他”字符的列。
2. 仅保留特殊列长度大于 1 的行。
3. 堆叠“特殊”和“其他列”以获得新列“y”

代码：

df["special"] = df["y"].apply(lambda x: [char for char in x.split(", ") if char in special])
df["other"] = df["y"].apply(lambda x: [char for char in x.split(", ") if char not in special])

output = df[df["special"].apply(len)>1]
output = output.drop("y", axis=1).set_index(["x", "z"]).stack().rename("y").reset_index().drop("level_2", axis=1)
output = output[output["y"].apply(len)>0][["x","y","z"]]
output["y"] = output["y"].apply(", ".join) #convert list back to string

输出：

>>> output
   x        y  z
0  a  x, m, d  b
2  a  x, m, d  t
3  a     g, i  t
4  a     x, m  k
5  a        i  k

输入：

df = pd.DataFrame({"x": ["a", "a", "a", "a"],
                   "y": ["x, m, d", "d", "x, m, d, g, i", "x, m, i"],
                   "z": ["b", "c", "t", "k"]
                   })
special = ["x", "m", "d"]

>>> df
   x              y  z
0  a        x, m, d  b
1  a              d  c
2  a  x, m, d, g, i  t
3  a        x, m, i  k

编辑

为多个列表合并 50% 规则（在问题发生重大变化后）

代码：

df["A1"] = df["y"].apply(lambda x: [char for char in x.split(", ") if char in A1])
df["A2"] = df["y"].apply(lambda x: [char for char in x.split(", ") if char in A2])
df["A3"] = df["y"].apply(lambda x: [char for char in x.split(", ") if char in A3])

output = df.drop("y", axis=1).set_index(["x", "z"]).stack().rename("y")
output = output[output.apply(lambda x: (all(c in A1 for c in x) and len(x)>=0.5*len(A1)) or
                                        (all(c in A2 for c in x) and len(x)>=0.5*len(A2)) or
                                        (all(c in A3 for c in x) and len(x)>=0.5*len(A3)))]
output.droplevel(2).reset_index()[["x", "y", "z"]]

输出：

>>> output
   x             y  z
0  a     [x, m, d]  b
1  a     [x, m, d]  t
2  a        [x, m]  k
3  a  [dd, cc, aa]  j
4  a      [yy, ll]  j

输入：

df = pd.DataFrame({"x": ["a", "a", "a", "a", "a"],
                   "y": ["x, m, d", "d", "x, m, d, dd, cc", "x, m, dd", "yy, ll, dd, cc, aa"],
                   "z": ["b", "c", "t", "k", "j"]
                   })

A1 = ["x", "m", "d"]
A2 = ["dd", "cc", "aa", "bb", "tt"]
A3 = ["kk", "yy", "ll", "ss"]