f***a 发帖数: 329 | 1 Hierarchical model:
y_i given mu_i is independent pois(mu_i) for i=1,...,n;
f(mu_1, ..., mu_n) is multivariate distribution with parameters (beta_1, ...
, beta_n);
(y_i 's are observed data and mu_i 's are latent variables here)
I ran random walk metropolis-hasting and got posteriors for mu_i and beta_i
which all converge well. But when extending model to beta_i=b0+b1_i*x_i (x_i
's are "location" of data points and known and b0 is known), I tried to use
the same algorithm but failed to get good |
|