Skip to contents

shinymrp 0.8.0

New features

  • Handle continuous outcome measure (#66)

Bugfixes

  • Fix incorrect order of levels in demographic distribution and estimate plots (#78)
  • Fix model compilation error caused by empty prior input (#77)

shinymrp 0.7.1

Bugfixes

  • Fix incorrect estimates caused by level mismatch between sample and poststratification data (#63)
  • Fix model compilation errors caused by including multiple varying slopes with the same categorical variable (#58)
  • Fix model compilation errors caused by interaction terms involving categorical variables included as fixed effects (#57)
  • Bugfix for static LOO-CV table (#59)
  • Fix error in “View results” tab caused by switching interface version (#62)
  • Fix consecutive model tab removal error (#61)
  • Bugfix for poststratification table upload for general time-varying data (#60)

shinymrp 0.7.0

New features

  • Upgrade UI to Boostrap 5 standard (#42)
  • Add diagnostics sections for model fitting and comparison (#40)
  • Add poststratification table upload (#54)
  • Add indicator for interface version/use case (#41)
  • Add download button for preprocessed data (#46)
  • Add back button and update use case description in the homepage ($48)
  • Add reference levels in estimated parameter table (#49)

Bugfixes

  • Suppress console output from compilation of standalone generated quantities code (#39)
  • Fix inconsistent ggplot styling on different machines (#28)
  • Prevent crashing when errors occur during LOO-CV (#51)
  • Omit interaction between nested variables (#50)

Other changes

  • Add documentation about input data requirements to in-app user guide (#31)
  • Notify users of the omission of variables with only one level (#29)
  • Minor UI improvements (#47)

shinymrp 0.6.0

New features

  • Extend ACS-data-linking to county and state (#19)
  • Add data linking section to data upload tab (#34)
  • Add two new interface versions for general use cases (#35)

Bugfixes

  • Fix demographic distribution bar plot error (#27)

Other changes

  • Make fit object self-contained by adding preprocessed data, poststratification table, and metadata (#43)
  • Change preprocessing code to include all tracts instead of only the ones in the urbanicity and ADI datasets (#24)
  • Fix letter casing of graph labels for counties and state (#44)

shinymrp 0.5.0

New features

  • Allow users to run poststratification separately (#20)
  • Significantly speed up poststratification by running it through Stan (#21)

Bugfixes

  • Fix errors thrown by renderPlot when users switch interface version (#25)
  • Conditional display of sections in table of estimates (#23)

Other changes

  • Speed up fit object uploading and downloading with qs

shinymrp 0.4.0

New features

  • Allow fixed effects of categorical variables
  • Filter effects compatible with structured prior
  • Add prior syntax validation and warning
  • Replace Learn > Interface with a popup “Guide” dialog

Bugfixes

  • Fix error that occurs when you run model fitting without including any effects
  • Change confidence interval of posterior draw summary from 90 to 95

Other changes

  • Change variable select input update logic
  • Minor UI changes

shinymrp 0.3.1

New features

  • Add structure prior for interaction between sex and categorical variables
  • Prevent a categorical from being included as both a fixed effect and a varying effect
  • Add C++ toolchain checking and CmdStan installation
  • Add model comparison to web version

Bugfixes

  • Fix arrow of summary HTML element not rendered on Window
  • Fix missing covariates when using example individual-level cross-sectional data
  • Variable select inputs persist after users navigate to another tab
  • Fix error in model comparison tab when users change interface version

Other changes

  • Restrict categorical varibles to varying effects in model specification
  • Explicitly set ggplot font to ensure consistency across platforms
  • Update Learn > Interface

shinymrp 0.3.0

New features

  • Revamp model specification UI and Stan-code generating backend
  • Add interaction terms to model specification
  • Add structure prior
  • Add random sampling for unseen levels

Other changes

  • Prevent users from proceeding to “Analyze” tab without selecting an interface version
  • Update default priors
  • Update Learn > Interface

shinymrp 0.2.0

New features

  • Allow user to save and upload model fits
  • Add buttons that allow user to use example data and model fits

Bugfixes

  • Fix model ordering in the LOO comparison table

Other changes

  • Feedback link takes users to a Google Form instead of GitHub Issue page

shinymrp 0.1.0

Beta release