otpod is a module for OpenTURNS.
This package enables to build Probability of Detection (POD) curves from Non Destructive Test. The curves are built using parametric models : univariate linear regression, quantile regression, kriging and polynomial chaos. Analysis can be run in order to test the linear regression hypothesis.
PoD can be built from a set of data or directly from a given physical model that simulate the Non Destructive Test. In this case, the design of experiments is defined iteratively.
Sensitivity analysis can be also be performed. The aggregated Sobol indices are available as well as the perturbation law indices.
This module is developped in python using several external modules :
- openturns = 1.11
- statsmodels >= 0.6
- numpy >= 1.10
- sklearn >= 0.17
- matplotlib >= 1.5
- scipy >= 0.17
- logging >= 0.5
- decorator >= 4.0
In a terminal, type in :
$ python setup.py install $ python setup.py install --user
Add the user option to install without administrator rights.
Or you can install the module using Anaconda repository.
$ conda install -c conda-forge otpod
The documentation with examples is available here.
Test are available in the 'test' directory. They can be launched with pytest and the following command in a terminal in the otpod directory:
$ pytest test/