Stat-JR bug fixes and new features

If you think you have found a new bug (we report known bugs below), then please report it using our Bug Report Form.

Stat-JR 1.0.5

21-Nov-2017

Move settings screen into drop-down dialogue box instead of separate page
Fix dataset download in TREE
Start directly in template run page when opening TREE
Save all outputs, regardless of whether they are recognised when downloading template run from TREE
Only record completed template runs
Fix cases when workflow cannot be exported from TREE
Avoid converting from Unicode twice when rending HTML outputs in TREE
Include unrecognised files when saving model outputs from TREE
Fix "More" button in TREE to work with recent changes to numpy
Fix input string not changing when selecting previous model runs in TREE
Fix behaviour of "Save" button after editing input resources in TREE
Remove link from Stat-JR:TREE in toolbar
Allow changing the matplotlib default styling from TREE settings
Justify text within HTML deep-dynamic outputs
Re-write eBook editor in TREE to contain one text editor per page
Move load/edit eBook from index to run page in TREE
Load eBook information from ttl file, instead of requiring a second definition file
Add initial support for static files in the TREE eBook editor
Add support for adding/removing statics files from eBooks in TREE
Add option to TREE for clearing the current eBook
Add support for evaluating xpath in the TREE eBook writer
Add support for rendering outputs with only/except expressions in the TREE eBook writer
Add support for DEEP specific HTML wrappers to the TREE eBook writer
Add support for table precision options in the eBook writer
Allow MathML elements when editing eBooks
Use previously stored binding names when writing out eBooks
Allow larger eBooks to be loaded in the editor by increaseing the maximum request size
Add option for clearing the current eBook
Add button for converting current page from dynamic to static in the eBook writer
Allow the user to answer template questions when a response is not defined when loading eBooks into TREE
Allow the user to select a dataset when loading eBooks into TREE if it has been set to the empty string
Record eBook inputs that are changed when loaded into TREE
Fix rendering LEAF in Internet Explorer
Interpret row names as strings when extracting values from a table in LEAF
Add '%' syntax for TextFormat blocks in LEAF
Avoid error in LEAF if a procedure contains an input question
Allow selection of templates/dataset from a dialogue box with LEAF
Enable blockly's custom variable category and support the Change By block
Add "Preview eBook" button to LEAF to hide outputs not created from a "Show" block
Allow specification of precision in tables
Improve Unicode support in eBook HTML
Improve print view in DEEP
Allow SVG images to be rendered in DEEP
Fix DEEP status display if a workflow is resumed
Only use text portion of headers when constructing the outline tree in DEEP
Fix some infinite looping bugs when running workflows in DEEP
Rework workflow input system
When reaching the end of a workflow in DEEP restart it
Display answers instead of retaining the question boxes for workflows in DEEP
Improve alignment of tables in DEEP
Don't include document outline tree when printing
Save XML versions of SPSS outputs
Changed SPSS package to use one background instance of SPSS, instead of opening a new version for each template run
Improve CSS handling with SPSS outputs
Expose current data in script and LaTeX sections of SPSS templates
Save dta files containing the FP and RP parameter matrices for the MLwiN packages
Save a version of the estimates that include the V matrices when running the MLwiN_IGLS package
Increase logging when running scripts with the MLwiN packages
Write separate logs files for each script when calling MLwiN
Display MLwiN log files if model results are missing from MLwiN packages
Add extra MCMC estimation options to the MLwiN_MCMC packages
Add options for additional parameterisations in MLwiN_MCMC package
Add code to MLwiN_MCMC package to try to ensure starting covariance matrices are valid
Move full name into description and use positions in variable names for MLwiN parameter outputs to avoid running over character limit
Add option to create cross-validation dataset to MLwiN packages
Unify rendering of outputs between modules
Remove unused modules
Preserve minus sign for number from -1 to -0
Fix case where value labels are specified in the dta file, but do not exist
Attempt to improve unicode handling for dta files
Synchronise xpath evaluation across modules
Increase code sharing across modules
Share more CSS between modules
Add more CSS to support eBook content
Allow XMLOutput types in xml_render function.
Add implementation of string endsWith() function to Internet Explorer support
Remove "placeholder" attribute in inputs, as this causes the text to be lost in Internet Explorer (see https://connect.microsoft.com/IE/feedback/details/811408).
Only allow block-diagonal or full covariance matrices in complex level-1 templates
Allow variables that start with underscore to work in XYPlot template
Allow SVG images to be rendered and improve MIME types
Improve handling of variable removal in multi-select boxes
Fix case when string values may not be null-terminated in newer Stata formats
Add negative-binomial option to various templates
Fall back on Python implementations for various functions if gcc is not found
Allow default values to be specified for data column inputs
Update to use numpy.histogram in various places as scipy.stats.histogram has be deprecated
Try to prevent tables being split across pages when printing
Always use red/green/blue colouring for BGRD graph
Ignore deviance when deciding whether to halt estimation due to parameter values of infinity
Use the default colour cycling for plotting templates
Synchronise OpenBUGS and WinBUGS model packages
Use the last values, rather than the starting values again after each stage of estimation with the eStat engine
Update documentation
Switch from jQuery-treeview to jqTree for document outline in DEEP interface
Remove jQuery-cookie
Remove jQuery-textalign
Update Blockly
Update TinyMCE
Update jQuery
Update jQuery-UI
Update Bootstrap


Stat-JR 1.0.4

27-Jun-2016

Added Workflow application module
Allow subset of dataset to be specified in DataFileChooser
Update TinyMCE, Bootstrap, jQuery, jQuery-file-upload
Allow multiple-select boxes to work when there is only one item
Fix case where the system attempted to read resources before they were fully available
Improve passing message between the browser and back-end
Reduce the amount of output generated by the back-end
Add a setting to the configuration file to change the logging level
Prevent file-checking dialogue in DEEP from becoming too large to fit on the page
Prevent the final template questions in DEEP from being repeatedly asked
Fix "categorial" parameter in DataMatrix constructor
Share more components between the different Stat-JR modules
Add initial support for running and refering to Workflows from DEEP
Add log-determinant function to the provided statistics library
Add initial support for hglm R package
Reduce storage requirements for running mean/sd
Update generation of missing data summary information in TREE to take into account numpy changes
Enable DIC calculation for 2-level multiple membership template
Use the first ID column when naming residuals dataset linked to IDMatrix
Correctly identify variables with missing data and include the sd with residuals generated from IDMatrix
Fix rendering error due to changes in numpy when there is missing data
Enable outputs with extension .htm to be correctly imported
For SPSS package, save HTML versions of outputs and allow processing post-processing outputs if requested in the template
Add initial support for PSPP
Change command used for SPSS in Regression2 template from GENLIN to REGRESSION
Remove support for raw numpy arrays as output (used previously for aML)
Handle case where results text is unicode in eStat and CustomC
Use version of Python that is bundles with SPSS to run models, instead of calling SPSS directly
Changed Calculate template to always use floating-point division
Avoid OUTPUT EXPORT command in SPSS, as it doesn't work in some modes.
Initial support for exporting TREE run history into a Workflow
Make all previous resources available in the TREE eBook editor
Add recognition of negative-binomial models to MLwiN_MCMC package
Fix JavaScript error if datafiles names were numeric
Allow removing variable from multiple-select boxes if allow_cat is turned off
Allow output tables to be smaller if they contain fewer than twenty rows
Fix for inputs missing from settings screen after reloading packages
Make binomial template consistent by always interpreting the response as counts and converting as necessary
Correct class names in template files
Remove R_MASS packages as the functions used as included in base R
Fix specification of Poisson offsets and Binomial denominator in various templates
Fix formula generation in various R templates
Remove support for R packages where they aren't capable of supporting all combinations offered by the template
Remove "denomiator" question from 2LevelUnordered as none of the packages used support it
Add support for negative-binomial to templates that use INLA
Add "about" boxes to the various Stat-JR modules
Add prediction dataset outputs for MLwiN related packages
Simplify the questions in the Complex level-1 templates


Stat-JR 1.0.3

05-Nov-2015

Update jQuery, jQuery-UI, Bootstrap, jQuery-fileupload, jQuery-treeview, jQuery-xpath, jQuery-cookie, TinyMCE
Fix eStat executable to allow "Identify option to work correctly"
Correct conversion of exp(x) - 1 to expm1(x) when optimising algebra
Add initial support for provenence recording
Downcast integers greater than 4 bytes, and doubles greater than 8 bytes when saving
Handle quotes in Stata model outputs
Correct comparisons with None
Fix eStat executable to work with Mono
Display unloaded data in light shade in dataset selection box
When adding data use masked arrays
Allow editing of variable descriptions
Avoid duplication by storing shared components in a common directory
Allow resizing of data/summary tables
Use \bf for bold in LaTeX
Allow embedded images to appear in HTML outputs
Correct matrix orientation in AverageAndCorrelation template
Add initial support for reading Stata 14 files
Add DTA output for model estimation results
Add option for displaying unmonitored results in CustomC engine
Check data and templates specified in settings exist, if not create them
Combine data and summary pages, and allow editing dataset description
Add option for copying and unloading data with data editing page
Don't attempt to create summary statistics for non-numeric variables
Greatly increase speed when reading in DTA files
Add option to download data from data view page
Allow changing the value labels assigned to a variable, and don't assume these match the variable name
Allow adding/removing records with value labels
Allow changing variable names
Move "var" directory out of the application directory and into a user-specific location
Run Python scripts in seperate thread to improve responsiveness
Avoid generating duplicate code if the same ID variable is specified more than once in a template
Correct the model specification in some of the R templates
Add the compiler option to the settings so that the path no longer has to be specified externally
Improve support for running on machines where European decimal notation is set
Allow a file name to be specified when downloading eBook from TREE
When writing DTA files use empty string to represent missing
Allow selection from previous model runs when setting template inputs
Work around crash in Internet Explorer when resizing list boxes
If saving the current DTA format isn't supported attempt to use an older version
Add initial support for the Nimble R package (http://r-nimble.org/)
Fix possible error when saving value labels
Prevent outputs from previous runs appearing within TREE when a new run has been started
Fix "max" option in numeric inputs


Stat-JR 1.0.2

29-Sep-2014

Update jQuery, jQueryUI, Bootstrap, TinyMCE and included Python libraries
Fix crash when modelled columns contain missing data
Improve handling of missing data in graph templates
Add UUID validity check when loading eBooks
Add two-column and concertina element support to DEEP
Add quiz output type
Improve the ebook writing functionality
Fix loading template inputs from an external file
Save template inputs into seperate file when generated eBooks through TREE
Allow more than one dataset per region in generated eBooks
Allow selection of fixed inputs in generated eBooks
Allow the label displayed in inputs to differ from the submitted value
Add dataset summary and resources generated by DEEP to output list in eBook generator
Scale images within eBooks to improve fitting in columns, etc
Update included eBooks
Update generated MH steps to not update non-missing rows for missing data parameters
Fix crash when parameters are defined with names starting "mis", but aren't "miss", "misb", "miso" or "misu"
Fix DEEP inputs and display when only/except attributes are used
Fix occasional paging problems in DEEP
Save unrecognised Stata dataset versions as 115
Cache MCMC diagnostic graphs once they have been generated
Fix data-deep-except not working on columns in tabular data
Update license files
Add extra missing check to CustomC package
Update UUIDs in included eBooks to valid versions


Stat-JR 1.0.1

25-Mar-2014

Update jQuery, jQueryUI, jQuery File Upload, Bootstrap and jqGrid
In imputed datasets only merge back data that was previously missing
Fix setting initial values for parameter matrices in *BUGS packages
Add extraction and display of model fit statistic for INLA and CARBayes models
Allow "enter" key to generate new lines when editing script files
Check that data has finished loading before allowing it to be used
Fix crash when value labels contained single quotes
Improve tool-tip positioning in data summary
Improve Internet Explorer support in DEEP
Expand distributions supported for initial value generation
Improve input mechanism in DEEP
Fix handling of check-box list inputs
Fix categorical question in DEEP when the same template input names are used more than once on a page
Increase size of template/dataset dialogue boxes in TREE
Prevent long questions from making the input boxes too small in DEEP
Fix retrieving results from SPSS models
Fix algebra system to work correctly on systems when the decimal seperator is set to comma
Add initial support for outputs in subdirectories
Add support in DEEP summary tables for missing data
Add missing count, description and value label indicator to DEEP summary tables
Fix support for running models in Stan
Listwise delete rows containing missing values for models that don't handle them in eStat engine
Fix locking issue in templates that call other templates
Clean up HTML output types when rendering
Updated Tutorial and PlotsViaR eBooks
Marked executables as large address aware to enable access to more memory on 64-bit machines


Stat-JR 1.0.0

Dec-2013

Initial full release


Earlier beta versions

Stat-JR 0.2: Jan-2013

Parallel processing
Editing of input files produced by Stat-JR
eBook writer
CustomC estimation engine
Starting values for MCMC methods: improved facilities for users to specify starting values in MCMC estimation.
New link-up to myExperiment
More information at the front-end of webtest interface, including template and dataset descriptions on the main (opening) page
Multi-select 'cloud' terms, on main page of webtest interface, to aid template choice
New formatting of webtest interface to bring it in line with its sister interface, the DEEP eBook tool.

Stat-JR 0.1: May-2012

First beta release

Edit this page