search

danielcfurr / edstan

Other
8 stars 3 forks source link

Does edstan provide any plotting options? #17

Open sunilkalmadi opened 5 years ago

sunilkalmadi commented 5 years ago

First of all, thank you very much for great package ! Incredibly useful. I was wondering if there are any functions to generate trace plots or information plots for item sets (such as the ones provided in other packages like mirt). I understand with edstan we would have entire posterior distributions for parameters instead of point estimates. Can you provide some hints or divert me to any existing resources that would make such plots from edstan fits easier? Thanks a lot.

danielcfurr commented 5 years ago

Greetings! Thank you for your kind words. I'm always happy to hear from people using the package.

In edstan, there is a function stan_columns_plot() that may be used to summarize Rhat and number of effective samples (n_eff) for all parameters. This is what I use to infer whether the chains have converged (Rhat) and have provided sufficient information about the posterior (n_eff). This is in line with the sort of approach Andrew Gelman recommends, I think. I haven't written up anything like an information plot, I'm afraid.

rstan has several functions for diagnostic plots, including stan_trace() for trace plots.

On Tue, Sep 3, 2019 at 1:35 AM Sunil Kalmady Vasu notifications@github.com wrote:

First of all, thank you very much for great package ! Incredibly useful. I was wondering if there are any functions to generate trace plots or information plots for item sets (such as the ones provided in other packages like mirt). I understand with edstan we would have entire posterior distributions for parameters instead of point estimates. Can you provide some hints or divert me to any existing resources that would make such plots from edstan fits easier? Thanks a lot.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/danielcfurr/edstan/issues/17?email_source=notifications&email_token=ACJLJ37ZTHOZ7N5WBLAOWF3QHYOWHA5CNFSM4ITDPMB2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HI5ND4A, or mute the thread https://github.com/notifications/unsubscribe-auth/ACJLJ32QHSSMR4TYRAWP74LQHYOWHANCNFSM4ITDPMBQ .

sunilkalmadi commented 5 years ago

Thank you very much for the response, and for the information on diagnostic plots. But, I think I miscommunicated. I was talking about plots with respect to learned (posterior) parameters such as item response curve (theta vs probability of scored response for an item/step), test information curve (theta vs total expected score of test) etc. Also, it would be great to have function to generate predictive posterior distribution over a specified covariate matrix. Any directions towards these or hints on easiest way to get these from edstan fits? Thanks again!