When predictors are uncorrelated -- the 95% CI for the VIF includes infinity and the graph in the performance plots is strange.

set.seed(123)

# Number of observations
n <- 100

# Generate two uncorrelated predictors
x1 <- rnorm(n)
x2 <- rnorm(n)

# Generate noise
noise <- rnorm(n)

# Generate outcome as a linear combination of the predictors and noise
y <- 0.5*x1 + 0.5*x2 + noise

# Create a data frame
data <- data.frame(y = y, x1 = x1, x2 = x2)

# View the first few rows of the data frame
head(data)
#>            y          x1          x2
#> 1  1.5633692 -0.56047565 -0.71040656
#> 2  1.3257661 -0.23017749  0.25688371
#> 3  0.3908632  1.55870831 -0.24669188
#> 4  0.4046770  0.07050839 -0.34754260
#> 5 -0.8255054  0.12928774 -0.95161857
#> 6  0.3587717  1.71506499 -0.04502772

mod <- lm(data = data, y ~ x1 + x2)
mod |> performance::check_model()

mod |> performance::check_collinearity()
#> # Check for Multicollinearity
#> 
#> Low Correlation
#> 
#>  Term  VIF  VIF 95% CI Increased SE Tolerance Tolerance 95% CI
#>    x1 1.00 [1.00, Inf]         1.00      1.00     [0.00, 1.00]
#>    x2 1.00 [1.00, Inf]         1.00      1.00     [0.00, 1.00]

^{Created on 2023-07-04 with reprex v2.0.2}

Reproduce the example with lm(data = data, y ~ x1 + x2), then inspect performance::check_collinearity() and performance::check_model(), including the performance-plot graph. Compare the reported VIF confidence intervals and graph against the uncorrelated predictors; done means the infinity interval and strange graph behavior are handled consistently.