RobertKubinec
summary
Package ideal Stan implements hierarchical ideal‑point (IRT‑style) models for binary and ordinal data, built in Stan with variational Bayes and automatic identification tricks. It includes missing data inference (censoring/hurdle form), plotting, LOO support, and runs on the coffee example: ~11,000 raters × 429 products, flagging highly polarizing items (Starbucks Via, Kicking Horse) and taking on the order of hours with VB. Roadmap: CRAN, marginal likelihood, multidimensional models and more item‑level regressions.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:10Measurement models
- 1:34Ideal point models
- 3:17Ideal point model
- 4:52IRT model
- 6:33Bayesian models
- 8:40Example
- 9:56Trigger warning
- 10:25Visualization
- 11:15Coffee
- 11:49Missing Data
- 12:49Future Plans
related talks
21mTeaching Data Sharing through R Data Packages
19mCan I trust that package?
01h 05mThe Beman Project: bringing standard libraries to the next level
30mScalable Automatic Machine Learning with H2O
40mInstructLab: Applying Open Source Methods to Building and Trai...
28mA community based open source LLM and what it can do for your business
32mPanel: Integrating Legend, Bosque, and Morphir For Open Reg Tech through FINOS
01h 09mThe Beman Project: Bringing C++ Standard Libraries to the Next Level