Many thanks to @DrDan (https://community.alteryx.com/t5/user/viewprofilepage/user-id/35). He is amazing and solved my problem.
Here is DrDan's solution:
Since my last response I found a work around for the problem, which is really something that will happen for binary classification problems. Given you are running Alteryx on a Windows VM under OSX, you will have pretty slow I/O. The one to address this is to drag the Model Comparison macro onto the Alteryx canvas, right-click on the the macro to open it up, and then open the R Tool of the macro. When you do this you will see the R code that is used. Scroll to near the bottom of the R code, where you will find the following lines of code:
getPlots(simple_plot_data, limits, "lift", num_models)
getPlots(simple_plot_data, limits, "gain", num_models)
getPlots(simple_plot_data, limits, "pr", num_models)
getPlots(simple_plot_data, limits, "roc", num_models)
To this block of code, add the following:
getPlots(simple_plot_data, limits, "lift", num_models)
Sys.sleep(0.1)
getPlots(simple_plot_data, limits, "gain", num_models)
Sys.sleep(0.1)
getPlots(simple_plot_data, limits, "pr", num_models)
Sys.sleep(0.1)
getPlots(simple_plot_data, limits, "roc", num_models)
The call to Sys.sleep(0.1) will pause execution of the R script for a tenth of a second, which should be enough time from the prior graph to be read by Alteryx, before R attempts to send the next graph to Alteryx. After you have done this, run the macro workflow, and then save it to a new name. Use the modified macros in your workflow.
