The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

Most typical and most deviant

Ingo Rohlfing

2026-05-17

library(MMRcaseselection)

One can use four functions to choose four types of cases without classifying all cases as either typical or deviant. The most typical and most deviant cases are proposed by Seawright and Gerring (2008). The most typical case has the smallest residual of all cases. The most deviant case has the largest residual of all cases. The two functions most_typical() and most_deviant() work in the same way and show you the case with its residual. The input into the function is an lm object.

df <- lm(mpg ~ disp + wt, data = mtcars)
most_typical(df)
#> Merc 450SE 
#> 0.03421046
most_deviant(df)
#> Toyota Corolla 
#>        6.34844

The most deviant case does not distinguish between cases that have a large negative and a large positive residual. Cases with a negative residual are overpredicted because the predicted outcome is higher than the observed outcome. Cases with a positive residual are underpredicted because the predicted outcome is lower than the observed outcome. It might not matter whether a case is overpredicted or underpredicted because both subtypes of outliers can have the same type of deviance. However, one might be interested in knowing whether a case has a positive or negative residual and what the most overpredicted and underpredicted cases are. This is what the functions most_overpredicted() and most_underpredicted() achieve, each taking an lm object as input.

# largest positive residual
most_underpredicted(df)
#> Toyota Corolla 
#>        6.34844
# largest negative residual
most_overpredicted(df)
#> Ferrari Dino 
#>     -3.40868

The package does not include functions for plotting the cases. There are multiple, very useful packages such as the olsrr package that can be used for the easy visualization of residuals.

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.