plot_curve

hybrid_learning.experimentation.fuzzy_exp.fuzzy_exp_vis.plot_curve(metrics_pd, x='recall', y='precision', ax=None, variables=None, constraints=None, model_key='maskrcnn_box', logic_types=('boolean', 'lukasiewicz', 'product', 'goedel'), formulas=None, add_diagonal=None, add_value_hints=True, add_markers=False, legend_outer=True, title='ROC {model_key} ({variable})', figsize=(10, 10), formula_name_col='formula_attrs', to_pretty_names=None, verbose=False, lims=((0, 1.0), (0, 1.0)), save_as=None)[source]

Plot a curve of y against x varying the value of variables[0] given constraints. For each entry in logic_types and formulas, and for each value of entries in variables[1:], produce a new line.

To just get a normal plot, call plot_curve(x='x', y='y', variables=['x'], add_value_hints=False, add_diagonal=False, ...).

Parameters

  • x – column name of the x-values

  • y – column name of the y-values

  • variables – keys of variables in the DataFrame to vary over; different values for the first one are plotted into one curve, further variables will create new curves for different values

  • constraints (Optional[Dict[str, float]]) – mapping of keys of variables in the dataframe to a fixed value; used to filter the DataFrame

  • model_key (str) – shorthand for constraints={'model_key': model_key}

  • logic_types (Sequence[str]) – filter the DataFrame for any of these values of the "logic_type" column

  • formulas (Optional[List[str]]) – if given, filter the DataFrame for any of these values of the formula_name_col column

  • formula_name_col (str) – see formulas

  • add_value_hints (bool) – whether to add small text fields with the values of variable at each sample point

  • add_diagonal (Optional[bool]) – add a diagonal; set to False to disable; else set to 'start_at_one' (default for PR-curves) or 'start_at_zero' (default else)

  • legend_outer (bool) – whether to put the legend of plots outside of the plot

  • title (str) – figure or axis title; may contain formatting placeholders {model_key}, and {variable} (for variables[0])

  • to_pretty_names (Optional[Dict[str, str]]) – dictionary mapping values of model_key and metrics to pretty names

  • verbose (bool) – whether to also prettily display all values of the relevant columns of the DataFrame

  • lims (Tuple[Tuple[float, float], Tuple[float, float]]) – tuple of ((xmin, xmax), (ymin, ymax)) to set the xlim and ylim of the axes

  • save_as (Optional[str]) – if given, either directory or file path where to save this plot

  • metrics_pd (DataFrame) –

  • add_markers (bool) –

Returns

all plotted rows of the DataFrame