I have the following to model my univariate time series data: with pm.Model() as model: ℓ = pm.Gamma("ℓ", alpha=2, beta=1) η = pm.HalfCauchy("η", beta=5) cov = η**2 * pm.gp.cov.Matern52(1, ℓ) gp = pm.gp.Latent(cov_func=cov) f = gp.prior("f", X = train_x) σ = pm.HalfCauchy(…

@sachinruk , looks like a broadcasting error. Does it work if you try train_y = np.ones(48) instead of train_y = np.ones(48)[:, None]? I’ll be able to take a close look during the week

Shape of posterior predictive samples of GP

Questions

sachinruk November 10, 2019, 11:08pm 4

Perfect! That fixed the problem.

Topic		Replies	Views
Understanding shape of values returned by sample_posterior_predictive Questions	3	915	January 26, 2020
Sample_prior_predictive() failing based solely on `samples` parameter Questions	9	999	March 22, 2019
Sample_prior_predictive bug, gives wrong shape for dependent variable? Questions	2	458	July 29, 2019
Sample_ppc shape Questions	5	2687	July 18, 2018
Poisson likelihood over a Gaussian process Questions	4	1406	November 13, 2019

Shape of posterior predictive samples of GP

Related topics