The main improvement in this release of stargazer is that it expands the set of
models that stargazer can represent.
Specifically, several models from linearmodels
(including both panel and
instrumental variable estimators) are now supported. This enhancement was made
possible by a cleaner separation, within stargazer, between generic and
estimator-specific code. As a result, supporting new models, including from
other libraries, should be straighforward.
As a second improvement, the new Label
object can now be used to provide
text that will adapt automatically to different formats. It is already used
internally for some stargazer-provided strings, and can be used by users for
own strings, such as variable names, so that for instance they look nice both
in Jupyter notebook (HTML format) and within a LaTeX document.
Both these novelties are featured in the examples notebook. In addition, this release introduces a large number of enhancements and bug fixes.
- New
Label
object, for multi-format text - Display pseudo-R² when available (:issue:`91`)
- Display marginal effects with
MarginalEffects
(:issue:`86`) - Implement more general computation of residual standard error (:issue:`76`)
rename_covariates()
can now accept a callable (in alternative to a dict)- Support for models from linearmodels library (:issue:`26`)
- New
restrict=False
argument tocovariate_order()
allows appending covariates not specifically ordered
- Escaped dependent variable name in LaTeX output
covariate_order()
now raises an error if a covariate name is repeated
The following people contributed commits to this release:
- Pietro Battiston
- phipz
- Dirk Sliwka
- Jonas Skjold Raaschou-Pedersen
This release brings a large refactoring, deduplicating much code between HTML and LateX output and potentially simplifying the future support for other output formats.
In addition, it introduces a large number of enhancements and bug fixes.
Note that the project URL on GitHub has moved to https://github.com/StatsReporting/stargazer
- Support escaping LaTeX special characters in
render_latex()
(:issue:`60`) - Allow for custom spacing between coefficient rows (:issue:`66`)
- Support the result of
get_robustcov_results()
fromStatsmodels
- Implement more general computation of residual standard error (:issue:`76`)
- Support for statistical results of other models than just OLS
- Introduction of a
utils.Wrapper
class allowing easy customization of any given model (:issue:`71`) - Make
separators
argument tocustom_columns()
optional - Allow changing the name of the dependent variable (:issue:`93`)
- Support custom lines (also) in header
- Fix width of line above mode numbers (:issue:`75`)
- Fix display of dependent variable in LaTeX
- Fix rendering of
generate_additional_notes()
(:issue:`80`) - Escape special characters (also) in columns labels (:issue:`90`)
- Fix typo in
LineLocation
value
The following people contributed commits to this release:
- Pietro Battiston
- Christoph Semken
- Isaac Liu
- Leo Goldman
- Mathias Ruoss
- Alex Dong
This release introduced a large number of enhancements and bug fixes.
- Support for multiple different target variables
- Empty rows for spacing are now optional
- Enabled rich display for HTML tables inside Jupyter notebooks
- Support models with missing features, hence beyond OLS
- Displayed thousands separator in number of observations
- Added
table_label
attribute and hidden the label if empty - Added the
show_stars
attribute to not features significance signs - Added the
add_line()
method to insert custom content - Added first unit tests
- Honored behavior of several
show_*
attributes - Adapted creation of multiple columns in LaTeX
- Show dependent variable name in HTML
- Fixed representation of floating numbers, keeping constant the number of decimal digits
- Fixed the number of observations in models without constant (:issue:`32`)
- Fixed output of R^2 in LaTeX
- Workaround for statsmodels bug #6778 affecting output of f statistic (:issue:`33`)
The following people contributed patches to this release:
- Pietro Battiston
- Christoph Semken
- Brandon Zborowski
- Max Ghenis