Michael Levy

alpha doesn't work in geom_ellipse

3

``` r library(ggplot2) library(ggforce) ggplot() + geom_ellipse(aes(x0 = 0, y0 = 0, a = 10, b = 3, angle = 0), alpha = .1) + coord_fixed() ```  Created on...

It took me a few minutes to track down why my install was failing. Worked fine for me after installing imagemackick (via `brew update && brew install imagemagick` on OSX).

Hi guys :) It seems tibble 3.0 has broken a way that healthcareai uses recipes. Here's a simple reprex with the error when the outcome is absent and success when...

Double warning for all-unique, not-ignored columns

1

Only the warning from `prep_data` should be surfaced. ``` r suppressPackageStartupMessages(library(healthcareai)) data.frame(x = letters, y = 1:26) %>% prep_data() #> Warning in find_columns_to_ignore(d, c(rlang::quo_name(outcome), ignored)): #> The following column(s) have...

It works for Catalyst data where all date or datetime columns will end with DTS, but otherwise I don't think there's a way to declare or get datetime columns noticed,...

They seem to switch if newdata is provided or not: ``` r library(healthcareai) m

Handle logical outcomes

1

and outcomes with "TRUE" and "FALSE" entries as character or factor. May require a new custom recipe step.

Say you train a model on a subset of the data, or for whatever reason you burned a lot of cycles training a model and then you want a new...