Filtering rows based on multiple condition pandas

I have this table

And I want to filter the rows based on this condition

groupby(['id_employee', 'stat']) -> 
if stat == h:  take row with 'kind_sco' == 3 
elif stat == m:  take row with 'kind_sco' == 1 
else: take row with 'kind_sco' == 2

My current solution is like this

data[(data['stat']=='l') & (data['kind_sco']==2) |
   (data['stat']=='m') & (data['kind_sco']==1) |
   (data['stat']=='h') & (data['kind_sco']==3)]

so the expected output is like

But it will not effective if the condition is more complicated

What is the simpler solution? I already think to use groupby and filter() but dont know how to tweak it.

Thank you in advance.

Here the key is to use apply function, which passes each grouped data frame so that you can "apply" your logic to it (here filtering) and then return that data

import pandas as pd

data = [
    {"id_employee": "123", "stat": "h", "kind_sco": "1", "score": "93"},
    {"id_employee": "123", "stat": "h", "kind_sco": "2", "score": "76"},
    {"id_employee": "123", "stat": "h", "kind_sco": "3", "score": "12"},
    {"id_employee": "123", "stat": "h", "kind_sco": "4", "score": "91"},
    {"id_employee": "456", "stat": "m", "kind_sco": "1", "score": "64"},
    {"id_employee": "456", "stat": "m", "kind_sco": "2", "score": "60"},
    {"id_employee": "456", "stat": "m", "kind_sco": "3", "score": "56"},
    {"id_employee": "456", "stat": "m", "kind_sco": "4", "score": "90"},
    {"id_employee": "789", "stat": "l", "kind_sco": "1", "score": "90"},
    {"id_employee": "789", "stat": "l", "kind_sco": "2", "score": "76"},
    {"id_employee": "789", "stat": "l", "kind_sco": "3", "score": "89"},
    {"id_employee": "789", "stat": "l", "kind_sco": "4", "score": "45"},
]


df = pd.DataFrame(data)
df["kind_sco"] = df["kind_sco"].astype(int)  # convert to int for comparision

grouped = df.groupby(["id_employee", "stat"])


def filter_rows_in_each_group(d: pd.DataFrame):
    if (d["stat"] == "h").all():
        return d[d["kind_sco"] == 3]
    elif (d["stat"] == "m").all():
        return d[d["kind_sco"] == 1]
    else:
        return d[d["kind_sco"] == 2]


filtered_data = grouped.apply(filter_rows_in_each_group).reset_index(drop=True)
print(filtered_data)

The output is as expected

  id_employee stat  kind_sco score
0         123    h         3    12
1         456    m         1    64
2         789    l         2    76

Maybe you can make smth up with a mapper and apply ?

mapper = {
"l": data["kind_sco"].eq(2),
"m": data["kind_sco"].eq(1),
"h": data["kind_sco"].eq(3)
# add here more groups/conditions ..
}
out = (
data.groupby(["id_employee", "stat"], group_keys=False)
.apply(lambda g: g.loc[mapper.get(g.name[1])])
)

Output :

print(out)
id_employee stat  kind_sco  score
2          123    h         3     12
4          456    m         1     64
9          789    l         2     76