penalty.learning: Penalty learning data

In anujkhare/iregnet: Regularized interval regression

Description

A data set for testing the iregnet package. The problem is learning a penalty function for predicting the optimal number of peaks in ChIP-seq data.

Usage

data("penalty.learning")

Format

X.mat is a numeric matrix with 443 rows/observations and 22 columns/features. Each observation is a genomic segmentation problem, for which the features (quartiles, mean, etc) are easily computed. The y.mat is a numeric matrix with 2 columns. The first column is the minimal log penalty value that predicts a minimal error model (and -Inf if there is no lower limit). The second column is the maximum log penalty value that predicts a minimal error model (and Inf if there is no upper limit). The goal is to learn a penalty function f such that lower.limit < f(X.mat) < upper.limit.

Source

data(H3K4me3.PGP.immune.4608, package="PeakSegJoint")

References

http://jmlr.org/proceedings/papers/v28/hocking13.html

anujkhare/iregnet documentation built on Aug. 23, 2019, 8:24 p.m.