Пользовательская сортировка фрейма данных на основе нескольких столбцов

Пытаться:

# extract required areas
df["tmp"] = df.Location.str.extract(r"(Manhattan|Queens|Cleveland)")

# define order
df["tmp"] = pd.Categorical(
    df["tmp"], categories=["Manhattan", "Queens", "Cleveland"], ordered=True
)

# sort
df = df.sort_values(by=["State", "tmp"]).drop(columns = "tmp")

print(df)

Распечатки:

    Name     State             Location
0  Jamie  New York            Manhattan
5  Perry  New York     Uptown Manhattan
7  Julie  New York      Lower Manhattan
2   John  New York      Flushing Queens
6  Diane  New York     Briarwood Queens
1   Mary      Ohio   Downtown Cleveland
3    Sam      Ohio   Lakewood Cleveland
4  James      Ohio  Beachwood Cleveland