plotting of predictions fails if no predictors (e.g., intercept only models)

[DominiqueMakowski]

m <- lm(qsec ~ 1, data = mtcars)

modelbased::estimate_prediction(m) |> 
  plot()
#> Error: No `by` variable was detected, so nothing to put in the x-axis.

^{Created on 2026-04-08 with reprex v2.1.1}

Not sure what's the best solution here:

• Throw more informative error message
• Produce histogram / density plot (check_predictions() for Bayesian models should use our own prediction/plotting capabilities performance#897) instead
• The plotting of predictions should always produce distribution plots

[DominiqueMakowski]

@strengejacke @mattansb @bwiernik what do you think?

[mattansb]

IMO, throw more informative error message.

[strengejacke]

Agree.