visualize_most_interesting_samples

hybrid_learning.experimentation.fuzzy_exp.fuzzy_exp_vis.visualize_most_interesting_samples(df, conf, num_samples=10, best=False, verbose=False, one_figure=False, fig_scale=3, save_as=None, **iterator_args)[source]

Plot the masks for most num_samples interesting samples side by side. “Interesting” here refers to those with the lowest mean_interesting value (mean of formula mask pixel values for pixels in which one of the input part masks exceeds a value).

Parameters

  • best (bool) – whether to plot the best (True) or the worst samples with respect to mean_interesting

  • one_figure (bool) – whether to plot all into one figure or into single ones

  • iterator_args – arguments to ResultsIterator

  • verbose (bool) – whether to print information about the plotted samples in a DataFrame

  • df (DataFrame) –

  • conf (dict) –

  • num_samples (int) –

  • fig_scale (float) –

  • save_as (Optional[str]) –

Returns

the sub-DataFrame with information about plotted samples

Return type

DataFrame