Error in model validation

[VishnuEriyat]

I was trying to validate a model I developed and the error encountered is as given below. I am following the Pmetrics manual for doing this, cannot find what is the problem in my data/model

Quitting from lines 8-11 [unnamed-chunk-38] (plotly.Rmd)

Quitting from lines 8-11 (plotly.Rmd)
Error in add_smooth():
! Regression failed due to unequal x (n = 194) and y (n = 0).
Backtrace:

1. res$op$plot(resid = T, icen = "median", pred.type = "pop", outeq = 1)
2. Pmetrics::add_smooth(., ci = ci, line = loessLine, method = "loess")
   Execution halted
   The system cannot find the file F:\rifam\Runs\171\outputs\NPAGreport.html.
   1 file(s) copied.

[mnneely]

Hmm, what does res$op$data look like? It seems that you are missing information so it can’t generate the regression. You can also try to plot without the regression.

res$op$plot(resid = T, icen = "median", pred.type = "pop", outeq = 1, line = list(loess = FALSE))

[VishnuEriyat]

I tried res$op$plot(resid = T, icen = "median", pred.type = "pop", outeq = 1, line = list(loess = FALSE)) and after loading the NPAG run files, This was the result. The res$op$data is also attached.

[mnneely]

OK, let’s try some more to isolate the problem.

res$op$plot(resid = T, icen = "median", pred.type = "pop", outeq = 1, line = list(loess = TRUE) #default, so line argument technically not required, but want to double check

res$op$plot(resid = T, icen = "median", pred.type = "pop", outeq = 1, line = list(loess = FALSE, lm = TRUE)) #suppress loess, add linear regression

[sumithmathew9]

When I perform External validation with cycles=0, obsSD, wd and wds are not generated. This may be the reason why we are getting the error
Error in add_smooth(., ci = ci, line = loessLine, method = "loess") :
Regression failed due to unequal x (n = 340) and y (n = 0).

Please see below exRes2$op$data

This error is not coming when we do NPAG run with cycles >0 and plot residuals

[VishnuEriyat]

The Loess was added and this was the result

Then when the linear regression was tried