objectif

Estimer la médiane (quantile 0.5) et autres quantiles.

code minimal

import statsmodels.api as sm
import numpy as np

rng = np.random.default_rng(0)
X = rng.normal(size=(100,1))
y = 2*X[:,0] + rng.normal(scale=0.5, size=100)
Xc = sm.add_constant(X)
mod = sm.QuantReg(y, Xc).fit(q=0.5)
print(len(mod.params) == 2)

utilisation

print(float(mod.params[1]) > 0)

variante(s) utile(s)

import statsmodels.api as sm
res = sm.QuantReg(y, Xc).fit(q=0.9)
print(res.q == 0.9)

notes

• Moins sensible aux outliers que OLS sur la médiane.