On this page we collect downloadable resources originating from the lab. Pretty much all of it relates to understanding, estimating, or manipulating personalized content in one way or another. Our most prominent software project right now is StreamingBandit (seeĀ which is maintain by Jules Kruijswijk (with help from the others of the team, primarily Robin van Emden). You can access the code here.


  • RStorm: package for the development of streaming processing bolts (aka Twitter’s Storm) in [R]. Grab it here.
  • StreamingBandit (work in progress): REST server for streaming contextual bandit policies. More info here.
  • A [R] package for fitting mixtures of logistic regression models in data streams. Here.
  • A bunch of python code for running agent based simulations of the Ultimatum game. Here.
  • Streaming approximations to the EM algorithm (by Lianne Ippel). Here.

And, more generally, our lab’s GitHub repository, here.

Data sets

  • Offline evaluation of the user of persuasive messages in e-commerce. Dataset collected through random allocation. Download the zip file containing the data set (3103598 rows) and README file here.

Miscellaneous [R] stuff

  • Cpp / R function for computing the Bayes optimal solution for a 2-arm Bernoulli bandit problem with finite horizon (might take a while for large horizons). See this post for information and getting started. You can download the C++ source, or the compiled version for direct use in R (note that you will need Rcpp).
  • [R] S4 Class for fitting finite mixture of logistic regression models in a data stream. Download the file here. And, see this post for an example of the usage (now obsolete and replaced by the resulting [R] package – here).
  • Two little [R] helper functions to “push” an array. Useful when maintaining data in a data stream (e.g., for implementing a moving average). Download here.