GAM describing the relationship between correlation and VIF thresholds

A fitted generalized additive model describing max_vif as a function of max_cor in experiment_cor_vs_vif.

Usage

data(gam_cor_to_vif)

Format

A gam object.

Source

Generated internally from experiment_cor_vs_vif.

Details

The model parameters (basis dimension k and weight exponent) were selected via optimization, filtering for models in the top 90\

The final model uses squared Jaccard similarity as weights to emphasize cases with high agreement between cor_select() and vif_select().

Model performance:

• Adjusted R-squared: 0.834

• Deviance explained: 83.4\

• Effective degrees of freedom for smooth: ~6

See also

Other experiments: experiment_adaptive_thresholds, experiment_cor_vs_vif, prediction_cor_to_vif

Examples

data(gam_cor_to_vif)
plot(gam_cor_to_vif, shade = TRUE)