Hello Pymc3 community,
Basically, when I model a time series using a simple autoregressive technique and a GaussianRandomWalk it fits the data perfectly, thus overfitting. My strategy was to fit to 4 years of data, and use their common prior to avoid this overfitting. However, this isn’t working, possibly because the model has too many parameters (it can be seen here: Combining AR/Negative Binomial with Gaussian Random Walk) . Is there another way to avoid overfitting in this scenario?
Thanks for your time!