您可以使用布尔掩码来查找最大值和先前值的位置：

# get max per row
max_val = df.max(axis=1)

# identify max and values after
m1 = df.eq(max_val, axis=0)
m2 = ~m1.cummax(axis=1)

# mask values after and get min per row
min_val = df.where(m1 | m2).min(axis=1)

# update DataFrame
df['Max'] = max_val
df['MinBefore'] = min_val

输出：

ab  bc  cd  ef  eh  ij  Max  MinBefore
0   1   2  12   4   6   7   12        1.0
1   4   5   6   1  10   4   10        1.0
2   5   6   9   1   2   4    9        5.0
3  12   1   3   1   2   4   12       12.0