At current, the r2() method for model objects classed c("mlm","lm") returns $R^2$ values for both outcomes separately and describes the proportion of variance in each of the separate outcomes.
There are some applications, in particular for relative importance analysis/dominance analysis, in which it is valuable to obtain one $R^2$ value that integrates across outcomes and provides a single value that describes the proportion of variance explained in the system of outcomes by the predictors.
Two recommended $R^2$ metrics are discussed by Azen and Budescu (2006) for use in dominance analysis that are derived from Van den Berg and Lewis (1988).
Would propose to add a r2_mlm() function that calls both of the recommended $R^2$ metrics for "mlm" classed model objects. I would also be happy to implement this function and submit a pull request if the maintainers are open to this additional set of metrics.
References
Azen, R., & Budescu, D. V. (2006). Comparing predictors in multivariate regression models: An extension of dominance analysis. Journal of Educational and Behavioral Statistics, 31(2), 157-180.
Van den Burg, W., & Lewis, C. (1988). Some properties of two measures of multivariate association. Psychometrika, 53, 109-122.
bwiernik
commented
4 months ago
At current, the
r2()method for model objects classedc("mlm","lm")returns $R^2$ values for both outcomes separately and describes the proportion of variance in each of the separate outcomes.There are some applications, in particular for relative importance analysis/dominance analysis, in which it is valuable to obtain one $R^2$ value that integrates across outcomes and provides a single value that describes the proportion of variance explained in the system of outcomes by the predictors.
Two recommended $R^2$ metrics are discussed by Azen and Budescu (2006) for use in dominance analysis that are derived from Van den Berg and Lewis (1988).
Would propose to add a
r2_mlm()function that calls both of the recommended $R^2$ metrics for "mlm" classed model objects. I would also be happy to implement this function and submit a pull request if the maintainers are open to this additional set of metrics.References
Azen, R., & Budescu, D. V. (2006). Comparing predictors in multivariate regression models: An extension of dominance analysis. Journal of Educational and Behavioral Statistics, 31(2), 157-180.
Van den Burg, W., & Lewis, C. (1988). Some properties of two measures of multivariate association. Psychometrika, 53, 109-122.