predict.StabilizedRegression

object of class 'StabilizedRegression'.

object

matrix or data.frame for which the response should
be predicted.

newdata

boolean. If <code>TRUE</code> the model SR
(pred) is used to predict, if <code>FALSE</code> the model SR is used.

predictive_model

additional arguments affecting the prediction produced.

Predict functions for 'StabilizedRegression' objects.

Contains an implementation of 'StabilizedRegression', a regression framework for heterogeneous data introduced in Pfister et al. (2019) <arXiv:1911.01850>. The procedure uses averaging to estimate a regression of a set of predictors X on a response variable Y by enforcing stability with respect to a given environment variable. The resulting regression leads to a variable selection procedure which allows to distinguish between stable and unstable predictors. The package further implements a visualization technique which illustrates the trade-off between stability and predictiveness of individual predictors.

predict.StabilizedRegression: predict function

Description

Usage

Arguments