URGENT - major bug: as default all rows with LIDV=0 dropped in ggPMX?

[biencan1]

Describe the bug I'm working on response data on a scale from 0 to 41 with 0 being complete suppression. Large proportion of the data has zero values. When generating a number of plots (individual plots, vpc, etc) I noticed that this data was not plotted. I did further investigation to the source and I compared the original data with the data in the controller and noticed that none of the datasets in the controller which includes LIDV contains the observation with zero values.

To Reproduce Comparison or original data vs data in the controler:

Expected behavior Example of individual plots in Monolix:

Example of a vpc:

Screenshots Same plot in ggPMX:

vpc in ggPMX:

Package and platform (please complete the following information):

Additional context Is a work around with ggPMX 1.0 possible? If so it's urgently needed. I suspect all of the diagnostic plots generated in ggPMX are affected.

[baltcir1]

Thanks, @biencan1, for raising this issue. Indeed, this was a "quick and dirty" fix of issues with log scale of some plots (eg. DV vs PRED). Basically, what we need here is rather a fix for the situation where the user wants to look at log scale and there are zeros. Can we just keep all data points (NOT filtering those with LIDV==0) and then see what kind of warning or error we get when we use log scales for dv_vs_pred plot?

Dear @tynesjo, would you be so kind and look quickly into this? It might be an easy fix. Please let us know if you have an appropriate example...

[biencan1]

@baltcir1 Yes, I see the issue... But I guess in a situation where you have data with zero values you wouldn't be generating plots on log scale?

[tynesjo]

@baltcir1 Yes, should be able to investigate during today or Monday

[mattfidler]

This should be fixed @biencan1