To reduce the amount of boilerplate code required for visualisations, it would be great to have in-built plotting functions added to the main modules in Effective Quadratures. For example Poly.plot()
and Subspaces.plot_zonotope
().
It would be important to implement the *args and/or **kwargs properly so the user can customise plots, and perhaps return ax
and fig
handles for later customisation.
Here is a running to-do list of plotting functions to be added:
Univariate and bivariate polynomials, including sigma bounds and quadrature points.
Dimension reduction - summary plots
Dimension reduction - 2D contours over subspace, zonotopes etc.
Polynomial trees - Partitions in feature space, graphviz of tree (already done).
Poly coefficients.
Sobol indices bar charts.
Index sets.
PDF’s of parameters.
Predictions vs truth for some points
Solver plots. e.g. convergence plots for iterative solvers, and regularisation path for elastic net.
I’ve made this post a wiki others should be able to add their own plotting ideas to the list.