A set of objects are contained here to easily facilitate the use of outcome transformations for modeling. For example, if there is a large amount of variability in the resampling results for the Kappa statistics, which lies between -1 and 1, assuming normality may produce posterior estimates outside of the natural bound. One way to solve this is to use a link function or assume a prior that is appropriately bounded. Another approach is to transform the outcome values prior to modeling using a Gaussian prior and reverse-transforming the posterior estimates prior to visualization and summarization. These object can help facilitate this last approach.
1 2 3 4 5 6 7 8 9
An object of class
list of length 2.
logit_trans object is useful for model
performance statistics bounds in zero and one, such as accuracy
or the area under the ROC curve.
inv_trans can be useful when the statistics
are right-skewed and strictly positive.
Fisher_trans was originally used for correlation statistics
but can be used here for an metrics falling between -1 and 1,
such as Kappa.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.