I’m trying to implement a RBF kernel for regression (similar to Kernel used in sklearn SVR). I want to capture non-linear relationships between the predictors and dependent variable. gamma = 1 with pm.Model() as model: x_i = pm.Normal("alpha", mu=0, sd=100, shape = X.shape[1]) w = pm.Normal…

I am no completely sure what is the implementation behind the kernel regression in sklearn, but I manage to make your example work: # bayesian rbf kernel with pm.Model() as model: # hyper-prior for ls ls = pm.Beta("ls", alpha=1e-2, beta=1e-4) cov_func = pm.gp.cov.ExpQuad(X.shape[1], ls=…

Regression using RBF kernel

Questions

Nadheesh June 1, 2018, 9:49am 14

Ok, that make sense.

One last question. In GP, are we considering the noise or the covariance (or both) as the uncertainty of the predictions?

Topic		Replies	Views
Understanding the Predictive Covariance of Gaussian Processes Questions	1	459	June 23, 2018
Tutorial real work about simple and multiple linear regression bayesian Questions	13	2201	September 20, 2018
Gaussian Process Regression level 1 inference: Re-producing Mauna Loa CO2 Example with PyMC3 Questions	10	2244	August 16, 2017
Multiple (uncertain) function observations of the same Gaussian process Questions	15	6568	October 16, 2017
Random forest kernel Questions	1	1306	July 16, 2017

Regression using RBF kernel

Related topics