fn.not.zero: Penalized models with non-zero fn at penalty=0

In tdhock/aum: Area Under Minimum of False Positives and Negatives

Penalized models with non-zero fn at penalty=0

Description

Usually we assume that fn must be zero at penalty=0, but this is not always the case in real data/labels. For example in the PeakSegDisk model with penalty=0, there are peaks almost everywhere but if a positive label is too small or misplaced with respect to the detected peaks, then there can be false negatives.

Usage

data("fn.not.zero")

Format

A data frame with 156 observations on the following 5 variables.

example

a character vector

min.lambda

a numeric vector

max.lambda

a numeric vector

fp

a numeric vector

fn

a numeric vector

Source

https://github.com/tdhock/feature-learning-benchmark

tdhock/aum documentation built on Oct. 26, 2024, 5:39 a.m.