Diagnostics

Diagnostics

The Diagnostics node runs simulation-based residual checks for a fitted linear or generalised linear model using the R DHARMa package. It produces a diagnostics plot that helps assess model fit and common assumption issues.

What it does

• Retrieves a stored model from an upstream Linear Model or GLM node
• Runs simulateResiduals() from DHARMa
• Renders the standard DHARMa diagnostics plot
• Displays the generated plot in the Output tab with zoom and download controls

How to use it

1. Fit a model - add a Linear Model or GLM node, enter a formula, and tick “Output to storr”
2. Connect the model - drag an edge from the model node to the Diagnostics node
3. Click Run - DHARMa residual simulation is run on the server and the diagnostics plot is returned

Configuration

Setting	Required	Description
Model connection	Yes	A Linear Model or GLM node with “Output to storr” enabled
Plot type	Yes	Choose between combined QQ+Residuals, QQ only, or Residuals only
Comment	No	Annotation for generated R code

Output

Displays the DHARMa diagnostics plot image for the connected model.

The plot can be:

• Zoomed for closer inspection in a modal view
• Downloaded as a PNG file for reports or sharing

Generated R code

library(DHARMa)
model_lm_dharma <- simulateResiduals(model_lm)
plot(model_lm_dharma)

Alternative plot calls:

plotQQunif(model_lm_dharma)
plotResiduals(model_lm_dharma)

Tips

• If the node errors with a package message, install DHARMa on the backend R environment.
• This node expects an lm or glmmTMB object from a model node, not a dataframe.
• Re-run diagnostics whenever you change the upstream model formula or dataset.