software to generate a reproducible container with a battery of experiments
Clone or download
Pull request Compare This branch is 17 commits behind expfactory:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

The Experiment Factory

DOI DOI Gitter chat


See our documentation for getting started. If you are new to containers, read our background or paper first. If you want a more guided entry, see the detailed start

The Experiment Factory is software to create a reproducible container that you can easily customize to deploy a set of web-based experiments.


If the Experiment Factory is useful to you, please cite the paper to support the software and open source development.

Sochat, (2018). The Experiment Factory: Reproducible Experiment Containers. Journal of Open Source Software, 3(22), 521,


We have many ways to contribute, and will briefly provide resources here to get you started.

How to Contribute

If you are a developer interested in working on the Experiment Factory software you should read out contributing guidelines for details. For contributing containers and experiments, see our user documentation. If you have any questions, please don't hesitate to ask a question.

Code of Conduct

It's important to treat one another with respect and maintain a fun and respectful environment for the open source community. Toward this aim, we ask that you review our code of conduct


It's predecessor at was never able to open up to the public, and this went against the original goal of the software. Further, the badly needed functionality to serve a local battery was poorly met with expfactory-python as time progressed and dependencies changes.

This version is agnostic to the underlying driver of the experiments, and provides reproducible, instantly deployable "container" experiments. What does that mean?

  • You obtain (or build) one container, a battery of experiments.
  • You (optionally) customize it
    • custom variables (e.g., a study identifier) and configurations go into the build recipe
    • you can choose to use your own database (default output is flat files)
    • other options are available at runtime
  • The container can be easily shared.
  • You run the container, optionally specifying a subset and ordering, and collect your results

If you build on Docker Hub anyone else can then pull and use your exact container to collect their own results. It is exact down to the file hash.

Experiment Library

The experiments themselves are now maintained under expfactory-experiments, official submissions to be found by expfactory can be added to the library (under development) to be tested that they meet minimum requirements.