This is a draft pull request for some of the modifications discussed here which will allow for different density estimation techniques.
Among the major modifications:
[ ] sample.evaluate_pdf and sample.evaluate_marginal_pdf have been altered to evaluate different pdfs. Previously, prob_type "kde" only evaluated marginal clustered kdes. This has been changed to "kde_marginals". Now prob_type "kde" expects a scipy.stats.gaussian_kde object for evaluation. Other prob_types have been introduced.
[ ] New function "generate_densities" is introduced in calculateR. The goal of this function is to provide a framework to apply alternative methods for output density estimation and save these density objects for evaluation in the sample_set_base objects using their native prob_type and prob_parameters attributes. "generate_densities" will be a wrapper function and should be flexible enough for new methods of density estimation to be applied.
[ ] calculateR.invert will have an option to call generate_densities with using different kinds of methods.
This is a draft pull request for some of the modifications discussed here which will allow for different density estimation techniques.
Among the major modifications:
sample.evaluate_pdf
andsample.evaluate_marginal_pdf
have been altered to evaluate different pdfs. Previously, prob_type "kde" only evaluated marginal clustered kdes. This has been changed to "kde_marginals". Now prob_type "kde" expects a scipy.stats.gaussian_kde object for evaluation. Other prob_types have been introduced.calculateR
. The goal of this function is to provide a framework to apply alternative methods for output density estimation and save these density objects for evaluation in thesample_set_base
objects using their native prob_type and prob_parameters attributes. "generate_densities" will be a wrapper function and should be flexible enough for new methods of density estimation to be applied.calculateR.invert
will have an option to call generate_densities with using different kinds of methods.