screen: SuperLearner screening wrappers

In lendle/hdps: High-dimensional propensity score algorithm

Description

Functions to set up screening wrappers for SuperLearner

Usage

screen.names(names)

screen.excludenames(names)

Arguments

names

Names to be included or excluded These functions generate simple screening wrappers for SuperLearner to include or exclude variables based on names. This is is helpful because in order to use HDPS as a candidate in SuperLearner, you need to include the study outcome variable as a covariate. But to use a non-HDPS algorithm, (say a random forest on some specified set of covariates,) as a candidate as well, you want to make sure you're not adjusting for the outcome which is downstream from treatment on the causal pathway. See documentation for the SuperLearner package for more about screening algorithms.

Examples

screen.predefined <- screen.names(c("names", "of", "predefined",
"covariates", "that", "definitely", "dont", "include", "the", "outcome"))

screen.notoutcome <- screen.excludenames(c("outcome_variable_name",
"and", "other", "covariates", "to", "exclude"))

lendle/hdps documentation built on May 9, 2019, 8:34 a.m.