I am new to Bayesian modeling. I am running Bayesian regression model in R using brm function from brms library, which is powered by STAN. I have a data with 10 million records. I took 10% sample out of it so as to run no Prior bayesian on it, and get Betas and est error, which I plan to use as priors for rest of the data. My queries are:
- All the beta estimates I am getting are having 0 est. error and non zero coefficent value, hence the beta distribution would be for e.g N(0.7,0) which isnt a distribution. Hence I cant use these results as prior
I have tried multiple itereation with 10%, 20%, 5% data with different number of chains and itereations combinations but the results are almost same. What should be my approach? My objective is to get proper estimates of all the variables and I want to do it with Bayesian regression itself as a part of learning.
Am I missing something here? Please suggest. Thanks in advance for your help.