objectif
Limiter les valeurs extrêmes via coupure aux quantiles.
code minimal
import pandas as pd
s = pd.Series([1,2,3,100])
q1, q99 = s.quantile(0.01), s.quantile(0.99)
w = s.clip(lower=q1, upper=q99)
print(w.tolist())
utilisation
import pandas as pd
df = pd.DataFrame({"x":[1,2,3,100]})
capped = df["x"].clip(upper=df["x"].quantile(0.95))
print(capped.max())
variante(s) utile(s)
import pandas as pd
s = pd.Series([1,2,3,100])
iqr = s.quantile(0.75) - s.quantile(0.25)
print(round(iqr,2) >= 0.0)
notes
- Alternative: détection via IQR et filtrage des outliers.