objectif

Calculer ranks et pourcentages de rang par groupe.

code minimal

import pandas as pd

df = pd.DataFrame({"g":["a","a","b","b"], "x":[10,30,20,40]})
print(df.groupby("g")["x"].rank(pct=True).tolist())

utilisation

import pandas as pd

s = pd.Series([100,50,75])
print(s.rank(method="dense").tolist())

variante(s) utile(s)

import pandas as pd

df = pd.DataFrame({"g":["a","a","b","b"], "x":[10,30,20,40]})
pr = df.groupby("g")["x"].rank(pct=True)
print(pr.groupby(df["g"]).mean().round(2).to_dict())

notes

• pct=True donne le rang relatif (0..1).