postpi_relate: postpi_relate function models the the relationship between...
In SiruoWang/IAP: What the Package Does (One Line, Title Case)

Description Usage Arguments Value Examples

View source: R/postpi_relate.R

This function is required to take in a data set (i.e. testing set) containing only observed and predicted outcomes and name for observed outcomes. It relates observed and predicted continuous outcomes through a gamma function. For categorical outcomes, user inputs observed outcomes and the probabilities of predicted outcomes (i.e. probabilities for each predicted category). It related categorical data through a user defined maching learning method from the caret package. The default method is k-nearest neighbours.

1	postpi_relate(relation_dat, yobs, method_categorical = "knn")

`relation_dat`	testing set that contains observed outcomes and predicted outcomes (continuous data) or probabilities of predicted outcomes (categorical data)
`yobs`	name of the observed outcome in the testing set
`method_categorical`	Method to be passed to caret train function for categorical data

rel_model relationship model between observed outcomes and predicted outcomes/probabilities

data(RINdata,package="postpi")

testing    <- RINdata[1:2000,]
relation_dat   <- data.frame(actual = testing$actual, pred = testing$predictions)
relation_model <- postpi_relate(relation_dat,actual)