R Packages


An R package providing a modern re-implementation of the Super Learner algorithm for ensemble modeling and stacked regression based on machine learning pipelines.
Joint work with Jeremy Coyle, Ivana Malenica, and Oleg Sofrygin.
[Docs] | [GitHub]


An R package providing a general framework for the application of various cross-validation schemes to arbitrary functions, facilitating the extension of cross-validation procedures to numerous applications.
Joint work with Jeremy Coyle.
[Docs] | [GitHub] | [CRAN]


An R package providing a fast and efficient implementation of the Highly Adaptive Lasso (HAL), a nonparametric regression estimator with optimality guarantees useful in semiparametric inference.
Joint work with Jeremy Coyle.
[Docs] | [GitHub]


An R package implementing efficient targeted maximum likelihood estimators of the effects of stochastic interventions framed as additive shifts on the treatment scale. Estimators are provided for the effect of a single shift as well as for a grid of proposed shifts, along with working marginal structural models for performing variable importance analysis using the Targeted Learning framework.
Joint work with Jeremy Coyle and Mark van der Laan.
[Docs] | [GitHub]


An R package implementing several estimators for the natural (in)direct effect under stochastic interventions, whether in the form of incremental propensity score shifts or modified treatment policies. These estimators flexibly extend causal mediation analysis to settings involving stochastic interventions.
Joint work with Iván Díaz.
[Docs] | [GitHub]


An R package for efficient estimation of and nonparametric inference on the effects of stochastic interventions, including in settings with multi-stage sampling designs. This provides a flexible way to perform nonparametric variable importance analysis on continuous quantities using the Targeted Learning framework.
Joint work with David Benkeser.
[Docs] | [GitHub]


An R package providing facilities for estimation and inference in right-censored survival analysis settings with and without competing risks, including extensions for data-adaptive target parameters, using Targeted Learning.
Joint work with David Benkeser.
[Docs] | [GitHub] | [CRAN]


An R package implementing a framework for using Targeted Learning to assess evidence for differential methylation across the genome by estimating variable importance measures at the level of CpG sites and related functional units.
Joint work with Mark van der Laan and Alan Hubbard.
[Docs] | [GitHub] | [Bioconductor]


An R package implementing a set of techniques for discovering biomarkers from biological sequencing data using a combination of Targeted Learning and a generalization of moderated statistics for variance stabilization in finite samples.
Joint work with Alan Hubbard.
View the package documentation and related information [Docs] | [GitHub] | [Bioconductor] |


An R package for data-adaptive hypothesis testing in high-dimensional settings. The approach allows for effects to be discovered (“mined”) from data without loss of valid statistical inference using the framework of Targeted Learning.
Joint work with Weixin Cai and Alan Hubbard.
[GitHub] | [Bioconductor]


An R package housing Nima’s personal R toolbox, largely containing miscellaneous convenience functions written to make statistical computing for research easier.
[Docs] | [GitHub] | [CRAN]

Nima Hejazi
PhD Candidate in Biostatistics


adaptest is an R package for performing multiple hypothesis testing in problem settings commonly encountered in high-dimensional …

methyvim is an R package providing facilities for nonparametric analysis of differential methylation using variable importance …

origami is an R package that provides a general framework for the application of cross-validation schemes to particular functions. By …

biotmle is an R package facilitating biomarker discovery by generalizing moderated statistics for use with targeted estimators of …