Stanify FAQ

[hyunjimoon]

Q. What if I fail to properly install cmdstanpy in my computer?

Ask questions in stan discourse. They are very kind and will likely to help you. If you face 'PCH' error, try this remedy. I've been saved countless time from install_cmdstan(), rebuild_cmdstan() command.

Q. How can I know I successfully created stanfiles in my computer?

Change setting['model_name'], and re-run demo_draws2datadraws.py file. If a new folder with the changed model_name is created in stan_files folder, there you go! Please do read stanfiles as its blocks would help you understand modules of Bayesian inference. Refer to Michael Betancourt's Introduction to Stan, especially the part that explains computation happening in each block.

Q. What if I succeed in creating stanfiles, but fail to compile and run stanfiles in my computer?

Don't worry. Two .nc files for each synthetic data and estimated samples is stored in data folder under the same name of your setting['model_name']. Using this, explore the structure of InferenceData. Especially, memorizing how each character (group) interact to make a story (relations) will help you to understand the last topic our class, simulation-based calibration. Detailed reasons for recommending this structure is explained here.

groups: posterior, posterior_predictive, sample_stats, prior, observed_data
relations: posterior, sample_stats, log_likelihood, posterior_predictive, observed_data, constant_data, prior, sample_stats_prior, prior_predictive, predictions, predictions_constant_data