This project implements Sensitivity Analysis methods. It is based on OpenTURNS.
The simplest analysis is to visualize how the quantity of interest respond individually to each input parameter:
pairplot(sample, data)
This analysis can be completed by quantitative information. Variance-based analysis is commonly used:
s, st = sobol_saltelli(function, 1000, 3, [[-np.pi, -np.pi, -np.pi],
[np.pi, np.pi, np.pi]])
plot_indices([s, st])
It is possible to use a polar coordinate system:
plot_indices([s, st], polar=True)
In case of an already existing sample, one can use density based measures:
momi = moment_independent(X, Y)
delta = momi[2]['Delta']
plot_indices([delta])
This method use not only the variance but all the PDF in order to compute sensitivity information. Also, it does not require the use of any particlar sampling design.
The dependencies are:
- Python >= 2.7 or >= 3.3
- numpy >= 0.10
- scipy >= 0.15
- OpenTURNS >= 1.12
- matplotlib >= 1.5.3
Using the latest python version is prefered! Then to install:
git clone git@github.com:.../otsensitivity.git cd otsensitivity python setup.py install