Companion code for a paper showcasing property-based sensitivity analysis as a useful approach in model development and quality assurance.

Setup

Clone the necessary repositories

$ git clone https://github.com/ConnectedSystems/diagnostic-sa.git
$ git clone https://github.com/ConnectedSystems/SALib.git --branch radial-oat-method --single-branch salib-roat 

Create conda environment (the specification for Windows platforms is provided)

$ conda create --name <env> --file win_env.yml

Install as a locally editable package

$ cd diagnostic-sa
$ pip install -e .
$ cd ..
$ cd salib-roat
$ pip install -e .

Full instructions as above:

$ git clone https://github.com/ConnectedSystems/diagnostic-sa.git
$ git clone https://github.com/ConnectedSystems/SALib.git --branch radial-oat-method --single-branch salib-roat 

# Replace environment name with your own
$ conda create --name <env> --file win_env.yml

$ cd diagnostic-sa
$ pip install -e .
$ cd ../salib-roat
$ pip install -e .
$ cd ..

Structure

The main set of code is found in the diagnostic_sa directory.

These are:

• li_2010.py

  Example implementations of the linear function taken from Li et al., (2010)

• li_2010_diagnostic_morris_incorrect.py

Analysis of the incorrectly implemented model

• li_2010_diagnostic_morris_inactive.py

Analysis of the model with an inactive parameter

Tests intended for use with the pytest framework is found in tests/test_li.py.

These produce the error messages as shown in the paper.

Code and analysis from an earlier revision is included here as currently the Open Science Framework cannot associate multiple repositories associated with a project.

These produce the figures included in Appendix A.

The original repository may be found here.