Scoring method for number of reverse dependencies a package has

Score a package for the number of reverse dependencies it has; regularized Convert the number of reverse dependencies length(x) into a validation score [0,1] $$ 1 / (1 + exp(-0.5 * (sqrt(length(x)) + sqrt(5)))) $$

Usage

# S3 method for pkg_metric_reverse_dependencies
metric_score(x, ...)

Arguments

x

a pkg_metric_reverse_dependencies packge metric object

...

additional arguments unused

Value

numeric value between 1 (high number of reverse dependencies) and

0 (low number of reverse dependencies)

Details

The scoring function is the classic logistic curve $$ 1 / (1 + exp(-k(x-x[0])) $$ with a square root scale for the number of reverse dependencies \(x = sqrt(length(x))\), sigmoid midpoint is 5 reverse dependencies, ie. \(x[0] = sqrt(5)\), and logistic growth rate of \(k = 0.5\).

$$ 1 / (1 + -0.5 * exp(sqrt(length(x)) - sqrt(5))) $$

Examples

if (FALSE) metric_score(assess_reverse_dependencies(pkg_ref("riskmetric")))