nnpiv.rkhs.ApproxRKHSIVL2

class nnpiv.rkhs.ApproxRKHSIVL2(kernel_approx='nystrom', n_components=10, kernel='rbf', gamma=2, degree=3, coef0=1, kernel_params=None, delta_scale='auto', delta_exp='auto', alpha_scale='auto')[source]

Approximate RKHS IV estimator with L2 regularization.

Instrument projections are represented by thin-SVD range bases. The sample-space empirical-L2 equation is contracted through the learner thin SVD so its pseudoinverse truncation is retained in a feature-sized system.