# generate config
from spux.utils import shell
shell.importer('config.py')
# plotting class
from spux.plot.mpl import MatPlotLib
plot = MatPlotLib()
# plot datasets
plot.datasets()
# plot marginal prior distributions
plot.priors()
# plot error model distribution treating each dataset point as prediction
# and a random realization of parameters from prior distribution
from datasets import datasets
from error import h
for name, dataset in datasets.items():
dataset = dataset()
xs = dataset['x'].copy(deep=1)
for index in dataset.index:
xs.loc[index] = h.inverse(xs.loc[index])
dataset[r'$\xi$'] = xs
plot.errors(predictions=datasets)
# plot distributions for the initial model values
from initial import initial
plot.distributions(initial, suffix='-initial')
# report status
# generate config
from spux.utils import shell
shell.importer('config.py')
# plotting class
from spux.plot.mpl import MatPlotLib
from exact import exact
plot = MatPlotLib(exact=exact)
# plot dataset
plot.dataset()
# plot marginal prior distributions
plot.priors()
# plot marginal error model distributions
plot.errors()
# plot marginal prior distributions for the initial model values
from inputset import inputset
exact_initial = exact['predictions'].iloc[0]
plot.distributions(inputset['initial'],
samples={'exact': exact_initial},
suffix='-initial')
# generate report
from spux.report import generate
generate.report(authors=r'Jonas {\v S}ukys')
# load posterior samples from spux.io import loader samples, infos = loader.reconstruct() # burnin burnin = 150 # plotting class from spux.plot.mpl import MatPlotLib plot = MatPlotLib(samples, infos, burnin=burnin) # plot unsuccessful posteriors plot.unsuccessfuls() # plot resets of stuck chains plot.resets() # compute and report approximated maximum a posterior (MAP) parameters estimate plot.MAP() # plot samples plot.parameters() # plot evolution of likelihoods plot.likelihoods() # plot evolution of likelihood accuracies plot.accuracies() # plot evolution of likelihood particles plot.particles()
# === LOADING
from spux.io import loader
samples, infos = loader.reconstruct(timingsfiles=None)
# === RESULTS
# plotting class
from spux.plot.mpl import MatPlotLib
from exact import exact
plot = MatPlotLib(samples, infos, exact=exact)
# plot unsuccessful posteriors
plot.unsuccessfuls()
# plot samples
plot.parameters()
# plot evolution of likelihoods
plot.distances()
# plot evolution of acceptances
plot.acceptances()
# plot pairwise joint posterior distributions
plot.posteriors2d(suffix='-progress')
# plot pairwise joint posterior distribution for selected parameter pairs
plot.posterior2d('drift', 'volatility', suffix='-progress')
# === RESULTS
# === load results from spux.io import loader samples, infos = loader.reconstruct () # === plot # burnin sample batch burnin = 75 # plotting class from spux.plot.mpl import MatPlotLib from exact import exact plot = MatPlotLib (samples, infos, burnin = burnin, exact = exact) # plot unsuccessful posteriors plot.unsuccessfuls () # plot samples plot.parameters () # compute Bayesian model evidence (best with burnin removed) plot.evidence (burnin) # plot evolution of likelihoods plot.likelihoods () # plot evolution of likelihood accuracies plot.accuracies () # plot evolution of likelihood particles