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

hajasw May 10, 2018, 9:41pm 9

Thanks very much Junpeng!
pm.AR works as you said it would.
Now all my problems are my own fault.

Cheers,
Wayne

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