I’m trying to write a latent variable Gaussian process regression in which the GP varies over time, not over my exogenous variable values. I’ve attached the code. I’m first simulating Ys, by taking the median of fake data generated in sim_data.R and using single_y_gp_sim.stan, then attempting to re…

Ok, so that time I got it! I wrote a function to generate the fake eta values within stan, instead of that AR(1) process I had before, and the values generated didn’t look at all like what I was getting in the first few tries. So the alpha I was trying to recover was unrealistic. I attached the co…

Gaussian process not recovering alpha parameter in cov_exp_quad

Modeling

bbbales2 September 17, 2017, 4:39pm 6

I’m gonna think some more, but see if there is anything helpful in this thread: AR/Kalman-like models where the k-th estimate of a variable depends only on obs 1..k-1

Topic		Replies	Views
Gaussian process predictive inference memory pressure Modeling	2	109	February 19, 2024
Identification of Matrix Normal Distribution General	2	514	July 4, 2021
Sharing a to-me-neat parameterization for Gaussian Processes Modeling techniques , specification	2	971	April 10, 2021
Advice me on GP prediction Modeling rstan , fitting-issues , specification	6	449	October 8, 2021
Likely-silly question on Gaussian Processes: can/should they be scale()'d? Modeling specification	1	577	February 10, 2022

Gaussian process not recovering alpha parameter in cov_exp_quad

Related Topics