There is an error in the doc:
The doc claims the equivalence C = n_samples / alpha, yet we don't know which alpha.

• If it's Ridge's alpha, it seems to me that C = 1 / alpha
• if it's SGD's alpha, it seems to me that C = 1 / (alpha * n_samples)

In both cases, the doc is not correct.

A quick overview gave me the equivalence : 1 / C_svc ~ 1 / C_logistic ~ alpha_ridge
~ alpha_elastic_net * n_samples ~ alpha_lasso * n_samples ~ alpha_sgd * n_samples