The Highly Adaptive Lasso 🤠📿
Switch branches/tags
Nothing to show
Clone or download
nhejazi Merge pull request #37 from tlverse/reduce-basis
Reduce basis functions used
Latest commit 338c881 Oct 17, 2018


Travis-CI Build Status AppVeyor Build Status Coverage Status Project Status: WIP – Initial development is in progress, but there has not yet been a stable, usable release suitable for the public. License: GPL v3

Fast and scalable estimation procedure for the Highly Adaptive LASSO

Authors: Jeremy Coyle and Nima Hejazi

What’s hal9001?

hal9001 is an R package providing an implementation of the scalable Highly Adaptive LASSO (HAL), a nonparametric regression estimator that applies L1-regularized regression (i.e., the LASSO) to a design matrix composed of indicator functions corresponding to a set of covariates and interactions thereof. Recent theoretical results show that HAL is endowed with several important properties that make it optimally suited for the purpose of estimating highly complex functions, including for inference in problem settings where causal parameters are estimated via data-adaptive techniques (i.e., machine learning), as is the case with Targeted Minimum Loss-Based Estimation (TMLE). While it is certainly possible to implement HAL purely in R, the computationally intensive nature of the algorithm suggests that writing core routines in C++ (and making these available in R via the Rcpp framework) ought to provide significant efficiency gains. hal9001 is just such an implementation.

For detailed discussions of the Highly Adaptive LASSO estimator, the interested reader might consider consulting Benkeser and van der Laan (2016), van der Laan (2017a), and van der Laan (2017b).


You can install the development version of hal9001 from GitHub via devtools with

devtools::install_github("tlverse/hal9001", build_vignettes = FALSE)


If you encounter any bugs or have any specific feature requests, please file an issue.


This minimal example shows how to use hal9001 to obtain predictions based on the Highly Adaptive LASSO. For details on the properties of the estimator, the interested reader is referred to Benkeser and van der Laan (2016) and van der Laan (2017a).

# load the hal9001 package
#> Loading required package: Rcpp
#> hal9001 v0.2.0: The Scalable Highly Adaptive Lasso

# simulate data
n = 100
p = 3
x <- xmat <- matrix(rnorm(n * p), n, p)
y <- x[, 1] * sin(x[, 2]) + rnorm(n, mean = 0, sd = 0.2)

# fit the HAL regression
hal_fit <- fit_hal(X = x, Y = y)
#> [1] "Look Dave, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill, and think things over."
#>                   user.self sys.self elapsed user.child sys.child
#> design_matrix         0.002    0.000   0.003          0         0
#> remove_duplicates     0.004    0.000   0.004          0         0
#> lasso                 0.249    0.007   0.255          0         0
#> total                 0.255    0.007   0.262          0         0

# training sample prediction
preds <- predict(hal_fit, new_data = x)
mean(hal_mse <- (preds - y)^2)
#> [1] 0.009266171


hal9001 is the primary implementation of the Highly Adaptive LASSO, an nonparametric function estimation procedure with numerous optimality properties. While contributions are very welcome, we ask that interested contributors consult our contribution guidelines prior to submitting a pull request.


After using the hal9001 R package, please cite the following:

      author = {Coyle, Jeremy R and Hejazi, Nima S},
      title = {{hal9001}: The Scalable {Highly Adaptive LASSO}},
      year  = {2018},
      howpublished = {\url{}},
      url = {},
      doi = {DOI_TBD}


© 2017-2018 Jeremy R. Coyle & Nima S. Hejazi

The contents of this repository are distributed under the GPL-3 license. See file LICENSE for details.


Benkeser, David, and Mark J van der Laan. 2016. “The Highly Adaptive Lasso Estimator.” In 2016 IEEE International Conference on Data Science and Advanced Analytics (DSAA). IEEE.

van der Laan, Mark J. 2017a. “A Generally Efficient Targeted Minimum Loss Based Estimator Based on the Highly Adaptive Lasso.” The International Journal of Biostatistics. De Gruyter.

———. 2017b. “Finite Sample Inference for Targeted Learning.” ArXiv E-Prints.