This is the current version of the Online SuperLearner for Time-Series data R package. Note that this version is in active development, and considered to be pre-alpha software. Be very careful interpretting any results from this package.


  • Automatic optimal predictor ensembling via sequential cross-validatio.
  • Can be extended with several algorithms.
  • Has several pre-defined summary measures

Install the development version from GitHub:


For an example on how to run the OnlineSuperLearner, view the Jupyter notebook, or the R/OnlineSuperLearner.Simulation.R file. For a complete guide see the documentation.

You can also run the demos for the project. Run:

The algorithm syntax

  • Each entry should have an algorithm entry with the name of the algorithm (i.e., the class name), and could have algorithm_params which is a list of hyperparameters for the specific algorithm. Furthermore, it could have a params entry with two entries: nbins specifying the number of bins to use for the discretization step, and online a boolean specifying whether the algorithm should be treated as an online one.

The intervention syntax

You can specify interventions as follows:

where variable is the variable to perform the intervention on, when is when the intervention should take place (at t= 2 in this example) and what what the intervention should be (1 in this case, but this could e.g. also be 0).


  • View the issues page


Polley EC, van der Laan MJ (2010) Super Learner in Prediction. U.C. Berkeley Division of Biostatistics Working Paper Series. Paper 226.

van der Laan, M. J., Polley, E. C. and Hubbard, A. E. (2007) Super Learner. Statistical Applications of Genetics and Molecular Biology, 6, article 25.

van der Laan, M. J., & Rose, S. (2011). Targeted learning: causal inference for observational and experimental data. Springer Science & Business Media.

Benkeser, D., Ju, C., Lendle, S. D., & van der Laan, M. J. (2016). Online Cross-Validation-Based Ensemble Learning. U.C. Berkeley Division of Biostatistics Working Paper Series, Paper 355.