Is there a practical way to implement autocorrelation in a pymc3 model? What I want to implement is: x_0~normal(0,s^2) x_i~normal( ro*x_i-1 , (1-ro^2)*s^2 ) for i=1,2,3,…n-1 In my current application, n=106 I can’t figure out how to implement this type of model in pymc3. I am hoping there is…

What you are descripting is exactly a AR model. For example, the PyMC3 code similar to about would be something like: with pm.Model(): T = pm.Normal('T', 7.78, tau=0.667) Ro = expit(T) Omega = tt.sqrt(1.-Ro**2) lnSigmaRecruit = pm.Normal('lnSigmaRecruit', 2.5, tau=1) tauRecruit …

Autocorrelation in a model

Questions

junpenglao May 9, 2018, 7:20pm 2

There is AR(n) distribution for this purpose, see eg http://docs.pymc.io/notebooks/AR.html

Topic		Replies	Views
Diagnosing AR(1) model whose results are way off course Questions	4	668	April 28, 2020
Vectorized autoregressive model Questions	5	2347	January 2, 2020
Code conversion Bayesian Modeling and Computation in Python ch6. Help on AR(1) model implementation Questions	5	393	July 11, 2022
Question about adding ARIMA into pymc Development	1	791	March 23, 2019
AR model "too many indices for array" Questions	1	1784	January 18, 2019

Autocorrelation in a model

Related topics