fuzzy_exp_eval

Description

Helper functions to evaluate fuzzy logic experiments. Includes functions for loading experiment results, post-processing them, and to do further evaluations.

Classes

ResultsIterator

Allow to get all output infos and potentially additionally calculated values for one sample.

Functions

add_cols_to_metrics_pd(metrics_pd)	Add some standard aliases and derived values to outputs of get_metrics.
auc_for(metrics_pd, model_key, logic_type, ...)	Collect area under curve of x-y-plots for the given experiment series.
formula_to_display_name(formula[, max_len])	Return a possibly shortened title version of the given formula string.
gather_aucs(metrics_pd[, other_by, ...])	Gather area under curve values for several formulas.
gather_exp_stats(conf, iterator[, thresh, ...])	Create a DataFrame summary of statistics of the masks created during a fuzzy logic experiment.
gather_results([metrics_csv])	Gather and cache DataFrame containing experiment metric results.
get_caches(root, conf[, discard_non_existing])	Given root and config provide the cache arguments needed for load_orig_and_masks
get_exp_conf(sacred_logdir[, ...])	Get the sacred config dict for the experiment under sacred_logdir.
get_metrics(experiment_root, model_key, ...)	Given a sacred experiment_root, collect all experiment settings and results for the respective filters.
load_orig_and_masks(img_fn[, orig_size, ...])	Load the original image and masks at index image_id from configured cache directories.
recalc_formula_masks(conf, masks, ...[, ...])	Given a sacred experiment conf``ig, changes to constants, and predicate ``masks, calculate the values of the given additional_formulas.
summarize_exp_stats(df, conf[, verbose])	Gather mean, std, min, max for the experiment stats DataFrame into a summary DataFrame (with these columns).
to_formula_attrs(formula)	Extract key properties from the formula string.
to_formula_attrs_str(formula)	Turn formula in nice short string listing its attributes.
to_logic_dirs(experiment_root, model_key, split)	Get the sub-directories for logic experiments based on given information.