Variance Inflation Factors Tool
The Variance Inflation Factor (VIF) tool produces a coefficient summary report that includes either the variance inflation factor or a generalized version of the VIF (GVIF) for all variables except the model intercept (which always has a VIF or GVIF that equals one). The tool produces VIF values if none of the predictors are categorical variables with more than two levels and produces GVIF values otherwise. The "standardized" (G)VIF values correspond to (G)VIF^[1/(2*DF)].
This macro is not automatically installed with Alteryx Designer. To use this macro, download it from the Alteryx Analytics Gallery.
This tool can be applied to the model objects created by the Linear, Logistic, Count, and Gamma regression tools in cases where the model is estimated using open source R routines, but not Revo ScaleR methods, since the Revo ScaleR model objects do not contain the information needed to calculate (G)VIFs.
This tool requires no configuration.
The two outputs relay similar data tables with differences in formats. The tables are generated from data given by the output of the vif package (citation) in R. The tables consist of the following information:
|Variable||The name of the variable|
|GVIF/VIF||The increase in the variance of a regression coefficient due to its collinearity with the other variables. This value is greater than or equals to one with larger values denoting a higher collinearity between the vector and the columns of the design matrix for that variable's regression on the other covariates.|
|DF||Degrees of freedom (for categorical variables, one less than the number of levels).|
|Std_GVIF/Std_VIF||This gives how many times larger the given variable is than if it was orthogonal to the others. It is calculated as (G)VIF^[1/(2*DF)].|
Connect a Browse tool to each output anchor to view results.
- D anchor: Outputs a Data table.
- R anchor: Outputs a Report table. The VIF values are formatted to have five significant digits.