Hi, I encountered some problems when building a Gaussian proxy model, I’m sorry it seems silly but I still cannot solve it. My training data X is 88×2 matrix, Y is 88×400 matrix. def main(argv=None): with pm.Model() as gp_model: ls = pm.Normal('ls',mu = 0.2, sd = 1) cov = pm.gp…

I think the issue here is that Y has multiple observations. Unfortunately this isn’t supported in PyMC3. If you check the docstring for gp.Marginal.marginal_likelihood, it says: y: array-like Data that is the sum of the function with the GP prior and Gaussian noise. Must have shape (n, ).

The predict function of GP

Questions

Vincy_1008 June 11, 2021, 8:31am 5

Thank you very much for your answer, this is really bad news, hope to see the upgrade in PyMC4.

1 Like

Topic		Replies	Views
Multidimensional input using Gaussian Process Questions	6	4028	June 28, 2017
Multi-output gaussian processes Questions	13	5511	October 22, 2017
Question about "gp.predict": Gaussian Processes Questions	2	851	September 2, 2021
Dimensions for y variable in gp.marginal_likelihood Questions gaussian_process	4	601	November 29, 2021
Can you provide some examples of multivariate gaussian process models in pymc3 Questions	0	369	April 21, 2020

The predict function of GP

Related topics