breaking change: Naming conventions for prop() and df_stats() have been adjusted somewhat. While this has the potential to break old code that relied on the names, the new system is clearer, supports a wider range of applications, and fixes a bug.
A counts() method for factors has been added. This method includes counts of 0 for levels that do not occur.
Formulas in df_stats() are preprosessed so that I() is now seldom (if ever) needed when doing on-the-fly calculations.
Added n_not_missing() and n_total() similar to n_missing().
Added ci.mean(), ci.median(), ci.prop(), and ci.sd() for computing confidence intervals.
fit_distr_fun() which fits distribtuions using MASS::fitdistr() but returns a function (density, cdf, quantile, or random sampling) instead of just the parameters of the fit.Additional functions from mosaic have been moved here.
Improvements to df_stats() to handle a wider range of stat functions.
mosaic and ggformula have been moved here to support a better dependency structure among the Project MOSAIC packages. This package is unlikely to be needed in isolation, but functions from this package will be imported into the other packages as needed. Some will also be re-exported.