Modeling right censored poisson regression

[Pedro_Ribeiro]

@Bob_Carpenter

Not quite. The way you wrote it the first time is correct. An observation is either observed with a value in range (hence the Poisson likelihood) or its censored. If it’s censored at inv[n], then we only know that the value is greater than or equal to inv[n], which means we only want to include the CCDF.

Thank you very much for the reply.
From a previous post

Clarification about definition of CCDF

Hello In the old version of stan user Manual https://mc-stan.org/docs/2_20/functions-reference/cumulative-distribution-functions.html the complementary cumulative distribution function is defined as Pr[Y≥y]. However in the new version https://mc-stan.org/docs/functions-reference/conventions_for_probability_functions.html#cumulative-distribution-functions the complementary cumulative distribution function is defined as P[Y>y] Which one of these is actually implemented in Stan ? Does CCDF i…

I understood that CCDF is defined as P [Y>y] . How is the term P(Y=inv) accounted ?