Freedom251 › Statistical Analysis Software
hal9001
What hal9001 is
hal9001 is a free R package implementing scalable highly adaptive lasso regression, a nonparametric estimator for complex functional forms under global smoothness assumptions. It applies L1-regularized lasso regression to indicator-function terms for covariate supports and their interactions. The package constructs sparse matrices of basis functions from observed data and includes a custom Lasso implementation for predictor matrices made entirely of indicator functions; glmnet is invoked by default. Data can be supplied as an R matrix, numeric vector, or formula interface. The maker describes uses including nuisance-parameter estimation. hal9001 runs through R on Linux, macOS, and Windows desktop systems and is distributed under GPL-3. CRAN provides Windows binaries, macOS arm64 and x86_64 binaries, and source code, plus a reference manual and fitting vignette. Installation is available with install.packages("hal9001"), while a development build uses remotes::install_github(). The package requires R 3.1.0 or later and Rcpp, and imports packages including Matrix, glmnet, data.table, and stringr.
Who it suits
It suits R users working with nonparametric regression, complex functional forms, or nuisance-parameter estimation. It accepts matrix, vector, and formula inputs.
What it does well
- Free GPL-3 package
- Builds sparse basis-function matrices
- Handles indicator-function predictor matrices
- Supports Linux, macOS, and Windows
- CRAN binaries and documentation available
What to know first
- Requires R 3.1.0 or later
- Requires Rcpp and imported packages
- Desktop package rather than a standalone application
Verdict
hal9001 provides a focused implementation of highly adaptive lasso regression in R. Its sparse basis construction, multiple input interfaces, and CRAN distribution suit users who can work within its R dependencies.
Facts from hal9001's own pages
- Available builds
- CRAN lists Windows binaries, macOS binaries for arm64 and x86_64, and a source package. · cran.r-project.org, 12 Oct 2026
- Basis construction
- The package includes routines for constructing sparse matrices of basis functions from observed data. · cran.r-project.org, 12 Oct 2026
- Contributions
- The project welcomes contributions and directs prospective contributors to its contribution guidelines. · tlverse.org, 11 Oct 2026
- Data import formats
- R matrix, numeric vector, or formula interface · tlverse.org, 24 Sept 2026
- Dependencies
- The package depends on R version 3.1.0 or later and Rcpp, and imports Matrix, stats, utils, methods, assertthat, origami, glmnet, data.table, and stringr. · cran.r-project.org, 12 Oct 2026
- Deployment
- desktop · tlverse.org, 24 Sept 2026
- Development install
- The development version can be installed from the tlverse/hal9001 GitHub repository using remotes::install_github(). · tlverse.org, 11 Oct 2026
- Development version
- The development version can be installed from the tlverse/hal9001 GitHub repository using remotes::install_github(). · tlverse.org, 12 Oct 2026
- Documentation
- CRAN lists a reference manual and a vignette on fitting the highly adaptive lasso with hal9001. · cran.r-project.org, 12 Oct 2026
- Estimation
- HAL regression can estimate arbitrarily complex functional forms under global smoothness assumptions. · tlverse.org, 11 Oct 2026
- Free plan
- Yes · tlverse.org, 24 Sept 2026
- Installation
- The maker recommends installing the package from CRAN with install.packages("hal9001"). · tlverse.org, 12 Oct 2026
- Integrations
- CRAN lists glmnet among the package imports and SuperLearner among its suggested packages. · cran.r-project.org, 11 Oct 2026
- Intended use
- The maker describes HAL regression as a way to estimate complex functional forms under global smoothness assumptions, with applications to nuisance parameter estimation. · tlverse.org, 12 Oct 2026
- Lasso fitting
- Its lasso routines use glmnet by default and include a custom implementation tailored to predictors composed of indicator functions. · cran.r-project.org, 12 Oct 2026
- License
- The package is distributed under the GPL-3 license. · tlverse.org, 12 Oct 2026
- Maintainer
- Jeremy Coyle is listed as the package maintainer on CRAN. · cran.r-project.org, 11 Oct 2026
- Method
- It applies L1-regularized lasso regression to indicator functions for covariate supports and their interactions. · tlverse.org, 11 Oct 2026
- Model basis
- It applies L1-regularized lasso regression to indicator-function basis terms for covariates and their interactions. · tlverse.org, 12 Oct 2026
- Platforms
- CRAN lists Windows and macOS binary packages and a source package; the package depends on R. · cran.r-project.org, 11 Oct 2026
- Purpose
- hal9001 is an R package implementing the scalable highly adaptive lasso, a nonparametric regression estimator. · tlverse.org, 11 Oct 2026
- R dependency
- The package depends on R version 3.1.0 or later and Rcpp. · cran.r-project.org, 11 Oct 2026
- Regression models
- Yes · tlverse.org, 24 Sept 2026
- Scripting support
- Yes · tlverse.org, 24 Sept 2026
Where hal9001 stands on Freedom251
- Statistical Analysis Software#49 of 106