Features introduced in version 1.0.5 (November 2017)
New Features and major changes introduced in Stat-JR version 1.0.5
- Stat-JR is now built as a 64-bit application (users updating from a previous version will need to install a new C++ compiler)
- The default graph plotting style for Python based templates can now be set by the user
- The output styling for HTML objects is improved
- Provide fallbacks for statistical functions if a C++ compiler is not found
- SAA templates have been introduced to support the eBook grant.
- Settings are now contained in a dialogue box
- The application starts directly on the templates specification and execution page
- The following eBook writer enhancements have been implemented
- Load/Save eBook files directly
- Support loading and managing static images
- Preview xpath evaluation
- Support only/except attributes
- Support DEEP HTML wrappers
- Preview specified table precsion
- Allow MathML elements
- Allow unspecified answers to be completed when loading an eBook
- Allow selecting data/templates via dialogue boxes
- Add placeholder syntax for text formatting
- Add "preview eBook" option to hide outputs not displayed via "Show block"
- Allow precision specification in tables
- Improve print rendering
- Display answers for questions after they are filled in
- Bring back full V matrix to Stat-JR
- Improve logging
- Add additional estimation options for MCMC estimation
- Implement option to perform cross validation
- Improve call times
- Bring back XML representations of outputs
- Improve support for HTML outputs generated
New Statistical Analysis Assistant (SAA) features
In version 1.0.5 we have introduced for the first time SAAs to Stat-JR. These take the form of eBooks accessed via the DEEP interface and allow the user to select a dataset and a type of statistical analysis and through answering a limited numbers of questions about their dataset Stat-JR will then perform a full analysis of their data including context specific text describing how to interpret the tables and figures constructed as part of the analysis. More details of these features are available in the new SAA guide (see http://www.bristol.ac.uk/cmm/software/statjr/manuals/) that talks about how the system was developed and guides the reader through several SAAs.
The majority of SAAs use MLwiN as the estimation engine for performing model fitting thus marrying the best features of both our StatJR and MLwiN packages. When the SAA has been run the resulting analysis can be downloaded as a pdf file and examples from the guide can be viewed (see http://www.bristol.ac.uk/cmm/software/statjr/manuals/).
The SAA system is evolving and we would love feedback of what you like and what you don’t about it as we hope in the future to construct analyst specific SAAs which perform statistical analysis the way you would as an analyst do them. If you encounter any issues with the software, or have any comments please let us know via CMM support (see http://www.bristol.ac.uk/cmm/software/support/).