Analysing the output

Module for analysing the output of APEMoST

The output files are in an easy-to-process format, so there is no requirement to use a specific set of tools.

class pyapemost.analyse.VisitedAllPlotter(outputfiles_basename=u'', nlast=0)

Extends VisitedPlotter to put all of those plots into one file (named chain0.pdf).

@see VisitedPlotter

class pyapemost.analyse.VisitedAnalyser(nlast=0)

Internal, abstract class that has the visited points of chain 0 loaded.

You need to overwrite:

marginal_plot(param1, values1)
conditional_plot(param1, values1, param2, values2)

@param nlast: if 0, use all points; otherwise only the last nlast ones.

class pyapemost.analyse.VisitedPlotter(outputfiles_basename=u'', nlast=0)

Produces individual plots of visited points for each parameter, and each parameter pair.

@param outputfiles_basename: prefix of output files

The output files are named chain0-paramname1-paramname2.pdf and chain0-paramname.pdf

class pyapemost.analyse.VisitedWindow

Not implemented / tested yet.

Should give a window to watch the progress of the Markov Chain, similar to VisitedAllPlotter but continuously updating.

pyapemost.analyse.create_histogram(parameter_name, nbins=100, writeFile=True, skipfirst=0, truncate=False, smooth=False)

Returns a histogram and some statistics about this parameter.

@param writeFile: if true, write the histogram to paramname.histogram

pyapemost.analyse.create_histograms(**kwargs)

Runs create_histogram for all parameters and returns a dictionary of the results

pyapemost.analyse.model_probability(show=True)

Calculate the model probability.

@param show: If true, print_model_probability is called

pyapemost.analyse.print_model_probability(logprob)

Gives a nice overview of the model probability, allowing the practitioner to compare this model’s probability to others