joint: Joint measures of predictive accuracy In DudbridgeLab/multipred: Calculates observed and expected measures of predictive accuracy over multiple traits

Description

Calculates joint sensitivity, specificity, positive and negative predictive value, concordance and net benefit for a vector of predictors.

Usage

 ```1 2``` ```joint(x, y, thresh = NULL, target = NULL, target2 = NULL, sample = 0, prev = NULL) ```

Arguments

 `x` Matrix of predicted risks. Each row corresponds to an individual, each column to a trait. Each entry should be a risk between 0 and 1. `y` Matrix of traits. Each row corresponds to an individual, each column to a trait. Must contain binary events coded as 0 and 1. `thresh` Vector of risk thresholds. For each row of x, an event is predicted for each trait that exceeds the corresponding element of thresh. These predictions are then compared to the elements of y. `target` Target trait vector for evaluating joint measures of accuracy. If NULL the marginal measures are calculated across the distribution of trait vectors. `target2` Second target vector for calculating joint concordance. `sample` Number of random samples to draw when estimating concordance. Defaults to 0, in which case all pairs of individuals in x are considered. `prev` Population prevalence of `target`.

Details

Joint measures consider the prediction of all events within a fixed target vector. If `target` is NULL, which is the default, then the marginal measures are calculated across the distribution of trait vectors. Unlike classical sensitivity, specificity and concordance, the measures then depend upon the distribution of traits in the input data, which may differ from that in the population.

If `prev` is not specified, it is estimated from the `y` matrix, ignoring ascertainment.

DudbridgeLab/multipred documentation built on May 28, 2019, 12:37 p.m.