test_dynamic_leveling(data,
samples=2000,
cut_away=0,
trials=5,
characteristic=' mean value ' ,
final_value=False,
sum_stats=False,
create_png=False)
| source code
|
Plot leveling of the given characteristic.
data ... cost criteria in chronological order
(out of optimization, format: numpy.array)
samples ... maximum of samples considered for leveling
cut_away ... number of data points to omit (skip initial stage)
trials ... how many different trials are used
characteristic ... statistical property to investigate
final_value ... if True, outputs the statistical property
calculated over the complete data set
sum_stats ... if True, replace trials with their mean and std values
create_png ... if True, a .png-file is created (otherwise interactive use)
Plot the leveling of the given characteristic. X-axis represents the
number of used samples to achieve the corresponding value. If sum_stats
is True, the standard deviation and the mean value is calculated over all
trials for each number of used samples.
To generate several trials, data is split into multiple parts. Thus, the
current approach to generate different trials is only applicable for
Monte Carlo generated date (randomly and independent at each point).
TODO!!! To investigate the leveling for other kind of data a new approach
is necessary! Until that use trials = 1 in the accordant cases.
|