Changelog

Extensions:

Added a save_settings_option to ExPanD() defaulting to TRUE so that users can start the app without the settings dialog at the bottom (issue #8)

Minor Issues:

Switched back to lfe for prepare_regression_table() OLS regressions. lfeis back on CRAN and allows for multiple fixed effects and multi-level clustering. It is now being run with cmethod='reghdfe' to make clustered standard errors consistent with Stata’s reghdfe

Bug fixes:

Extensions:

Added a prepare_by_group_trend_graph() function to plot time trends by grouping variable
Included grouped time trend plots in ExPanD()

Bug fixes:

Fixed notebook code for sub sampling (failed on sampling variables containing NAs)
Fixed a small issue in the Notebook generation code for the time trend and quantile trend chunks
Fixed display of prepare_regression_table() for logit model by group tables with fixed effects

Minor Issues:

Replaced lfe with plm package for fixed effect models as lfe has been removed from CRAN
Removed wbstats from Suggests as it is currently not on CRAN
Removed tidyquant package from Suggests as it depends on a package that requires R >= 3.5.0
Spelling

Extensions:

Added a vignette to explain the notebook export option of ExPanD() and the resulting notebook

Bug fixes:

Fixed bug that stopped ExPanD() displays from rendering on user uploaded panel data

Minor Issues:

Updated World Bank data
Allowed ExPanD() to be called with ts_id == NULL and cs_id != NULL so that data frame variables can be specified as cross-sectional identifiers in the cross-sectional mode
Removed some artefacts in cross-sectional mode where cross-sectional and time series identifiers were still provided as variable options in the ExPanD() app
Included manual zip package import in NAMESPACE to make devtoools::check() happy on various CRAN systems
Changed shinyapps.io URLs in vignettes to plain HTTP as win-builder seems to have an issue with the SSL handshake of shinyapps.io

Extensions:

Modified ExPanD() to process cross-sectional data with useful defaults
Implemented an export option that allows users to download a zip file containing the data and a R notebook based variant of the ExPanD analysis (export_nb_option = TRUE)

Minor issues:

Added an option ‘binary’ to prepare_missing_values_graph() to visualize whether any values are missing
Typo fixes

Extensions:

Introduced html_block tag in components to allow users to add self-designed html content to ExPanD()
Added sample_selection, subset_factor, grouping and udvars variables to components in ExPanD() so that these parts can also be re-arranged or omitted
Allowed user defined variables to be built on analysis variables and used in simple_mode
Added new vignette to explain new features

Bug fixes:

Minor issues:

Extensions:

Removed the requirement for at least one non-numerical variable
Added by group violin plot
Included ! and is.na() as allowed functions for user defined variables
Introduced the option to change the order of reported components and to exclude selected components
Added clustered standard errors and fixed effects for logit models in prepare_regression_table()
Implemented binary response logit models in ExPanD()

Bug fixes:

Minor issues:

Removed the dependency with the CodeDepends package
Added rio package to the imports list
Added packages used by ExPanD() to NAMESPACE to make CRAN checks happy
UI cleanups
Fixed the definition of oint_ta in r3 dataset
Added a check to verify that ts_id is provided as an ordered vector
Fixed a typo in worldbank_var_def
Removed the drop_undersore parameter in prepare_regression_table() (no longer needed)
Work-around for special characters in stargazer column.labels
Marginally improved error handling for user provided data files
Allowed treat_outliers() to route parameters through to stats::quantile() (needed for type parameter)
Changed NA handling in prepare_graph type functions

ExPanDaR 0.5.3.9000 Unreleased