I’m have multiple time series of the same length with the same AR parameters. I’m struggling to figure out the shapes/dims situation in v4. Here is a toy example: import pandas as pd import pymc as pm import numpy as np import arviz as az def create_ar_series(length, rho, sigma): res = [] …

That’s because I missed a zero. That works! with pm.Model(coords={"time": df.index.values, "series": [0, 1]}) as m: rho = pm.Normal("rho") init = pm.Normal.dist() sigma = pm.HalfNormal('sigma') ar1 = pm.AR("ar1", rho=rho, sigma=sigma, …

AR dims/coords/shape in v4

Questions v5

cluhmann August 11, 2022, 2:36am 6

Hm, I suppose that makes sense. Then I would vote for 2 likelihoods if you need to get something running in the meantime. @RavinKumar might have further suggestions.

Topic		Replies	Views
AR model "too many indices for array" Questions	1	1784	January 18, 2019
PyMC3+ArviZ: improve your workflow with labeled coords and dims Sharing doc	20	5877	April 5, 2021
Reproducing PyMC3 AR(1) tutorial in PyMC4 Development	1	983	June 7, 2020
Matrix Multiplication With Multiple Dimensions in PYMC Model v5 modeling	4	1114	July 7, 2022
Time-series regression - shape error / Input dimension mis-match Questions time_series	8	1186	August 6, 2022

AR dims/coords/shape in v4

Related topics