Scientific models are formalized hypotheses that can make predictions about system behavior.
How can we evaluate the strength of a candidate model, or select the best model among several competitors?
Which model best predicts the experimental observations? (Animation by @malincse)
SciUnit enables the reproducible execution and visualization of data-driven unit tests for assessing model quality.
What if every competing model was transparently evaluated against a suite of unit data-driven tests?
SciUnit aims to make validation of scientific models against experimental data easy, transparent, and continuously integrated into the model development process. It is the scientific method for 21st century scientists.
|SciUnit, a Pythonic framework for data-driven unit testing that separates the testing interface from the model implementation, respecting the diversity of conventions for modeling and data collection.||SciDash, a web application to schedule tests, or deposit completed ones, and to review and visualize test results online. If SciUnit is unit testing for models, SciDash provides a service analogous to Travis CI or CircleCI.|
A sample of domain-specific projects that use SciUnit can be found here, including NeuronUnit for neurons and ion channels.