- Added the
ANOVA: beyond the p-valuevignette. plot.eda_npolandplot.eda_mean_sweepmethods now allow for n-waymargindiagnostic plots- Fixed bug in
eda_boxlswhereby thebaseargument was not passed toeda_re eda_npolnow allows for interactive effects via themax_orderargument.- Added
edtts4.2data pulled from table 4-2 of Exploring Data tables, Trends, and Shapes. - Added
feav13_1data pulled from table 13-1 of Fundamentals of Exploratory Analysis of Variance. eda_matlongcan now handle n-factors and not just a two-dimensional table.- Median polish article: added text clarifying Tukey's interaction term as being an approximation to interaction.
- Re-wrote
eda_shuffle(). It now can perform permutations on one or more columns of a dataframe. - Fixes variability decomposition labeling problem for objects of class
eda_polish(#30) - Fixes missing regression line in median polish diagnostic plot method (#29).
eda_rfsnow outputs the f-values, residuals and fit-minus-mean values as a dataframe (#24).- Fixed bug in
eda_rfs()where the quantile type was not properly specified (#25). - Added a new function,
eda_mdens(), that allows for any number of batches and not just two as is the case witheda_dens. - Added a warning to
eda_dens()whenp != 1andbwis manually specified, to remind users to adjustbwfor the transformed data scale. - Added a
kernel = "sliding"option toeda_dens()for a sliding-window density estimator (#28). - Fixed bug in
eda_qq()wherepchandsizearguments were ignored (#27).
- Added internal function
.eda_plot_xyused by many XY plotting functions. - Added internal function
.eda_plot_vardecompused to view decomposed n-way tables. - Added new datasets--mostly used with n-way table polishing functions.
- Added the log base argument
basetoeda_re. - Added an exploratory ANOVA decomposition set of functions. These include:
eda_mean_sweep: The main ANOVA decomposition functionplot.eda_mean_sweep: Plot method foreda_mean_sweepeda_anova_table: Generates traditional ANOVA table fromeda_mean_sweepoutput.
- Added
eda_npolmedian polish function that allows for n-way tables with more than two factors. - Added plot method
plot.eda_npolforeda_npolobjects. - Added a variability decomposition plot,
eda_vd, that can be used with both n-way tables an linear models. - Added
eda_matlongthat converts a 2-way table between long form and matrix form. Useful when running tables througheda_polandmedpolishfunctions. - Modified
plot.eda_lmmethod to now make use of the.eda_plot_xyinternal function. Function also adds araw_tickargument that allows for the display of the raw (untransformed) variables on a transformed axis. - Added a
raw_tickargument toeda_lmthat allows for the display of the raw (untransformed) variables on a transformed axis. - Minor changes to
eda_boxls,eda_unipow,eda_bipow,eda_pol,eda_sym,eda_sl,eda_rlineandeda_qqto reflect the
newbaseargument ineda_reand changes to internal plotting functions. - Minor changes to the plot methods
plot.eda_rlineandplot.eda_polish.
- Took out the normal and symmetry options from
eda_qqand created their own functions:eda_qqsymandeda_theo. The latter allows for theoretical distributions other than Normal. - Added the
eda_fvalfunction that generates f-values (probabilities) based on one of six algorithms - Added the
eda_theopanfunction that generates multi-panel theoretical QQ plots. - Added the
eda_qqpoolfunction that generates multi-panel pooled residual plots. - Added the
eda_shufflefunction that permutes values across groups. - Added the
eda_simfunction that simulates distributions given kurtosis and skewness. eda_slnow outputs Cleveland's spread-location plot as well as Tukey's spread-versus-level plot.- Added
eda_untiefunction that splits ties in a vector by nudging them by a small amount. - Fixed bug in
eda_rlinewhere incorrect residuals were returned. - Fixed margin width for
eda_jitterandeda_boxlsplots when in horizontal mode.
- Added the
eda_rfsfunction (the Cleveland residual-fit spread plot). - Added the
eda_violfunction that generates a violin plot. - Added the
eda_jitterfunction that generates a jitter plot from univariate datasets. - Added the
eda_qqmatfunction that generates a qq plot matrix. eda_lmnow outputs aneda_lmclass.eda_rlinenow outputsfitted.valuesclass.eda_lmnow acceptspoly = 0. This generates a flat line. Can be used withrobust = TRUE.- Following changes were made to the
eda_qqfunction:- It now generates different point symbols for values outside of the
innerregion - The
innerregion default is now 75%. - The outer bounds dashed line argument,
l.valwas removed (this to de-clutter the plot) - Removed the quantile-difference plot. Its interpretation is tricky when dealing with a multiplicative offset.
- It now generates different point symbols for values outside of the
- Added the quantile-difference plot to the
eda_qqfunction. - Removed redundant
outliersargument fromeda_boxls(#21).
- Added symmetry QQ option to
eda_qqfunction. - Ensure that all re-expressions default to Box-Cox method.
- Added polynomial option to
eda_lmfunction. - Added robust fitting option to
eda_lmfunction (usesMASS::rlm)
- Added
show.paroption toeda_qq - fixed issue #17.
- Re-wrote eda_rline function. This also fixed issue #15.
- Added the option to plot the density distribution alongside the Normal fit in
eda_normfit.
- Added a Normal QQ plot option to
eda_qq. - Added symmetrical Normal fit plot function
eda_normfit. - Updated eda_boxls aesthetics.
- Updated median polish diagnostic plot aesthetics.
- Introduces the median polish function
eda_pol. - Introduces the QQ and Tukey mean-difference plot
eda_qq. - Introduces the density plot function
eda_dens. - Adds re-expression parameters to
eda_lmvia the parameterspxandpy. - Adds
sdlabels to SD dashed lines ineda_lm. eda_lmwill now outputlmintercept and slope.- Adds plot method for
eda_rlineobject. - In
eda_reifp = 1, box-cox option is ignored. - Homogenized plot appearances.
- Added power parameter argument to
eda_boxls. - Added power parameter argument to
eda_sl. - Added plot option to
eda_sl.
- Initial release of
tukeyedar