ReliabilityModelPlot.ranking

static ReliabilityModelPlot.ranking(df_rel, names=None, figsize=None, top_n=None, title=None, ax=None)

Per-sample view — each prediction’s score with its uncertainty, colored by trust status.

The core “should I trust this one?” figure: samples are ranked by score as horizontal bars with the confidence interval as an error bar, colored green (reliable), amber (in-domain but undecided), or red (out-of-distribution) — so an over-confident score on an unfamiliar input stands out as a long red bar.

Parameters:

• df_rel (pd.DataFrame) – Output of ReliabilityModel.predict() (needs score, ci_low / ci_high, in_domain, reliable).

• names (list, optional) – Per-sample labels for the y-axis; row positions are used if None.

• figsize (tuple, optional) – Figure size (used only when ax is None); scales with the sample count by default.

• top_n (int, optional) – Show only the top_n highest-scoring samples; all are shown if None.

• title (str, optional) – Axes title.

• ax (matplotlib.axes.Axes, optional) – Axes to draw on; a new figure is created if None.

Returns:

• fig (matplotlib.figure.Figure) – The created (or parent) figure.

• ax (matplotlib.axes.Axes) – The axes drawn on.

Examples

ReliabilityModelPlot().ranking() is the per-sample view — should I trustthisone? Each prediction is a bar (score) with its uncertainty interval as an error bar, colored by trust status: green = reliable, amber = in-domain but undecided, red = out-of-distribution. An over-confident score on an unfamiliar input shows up as a long red bar.

import aaanalysis as aa
import numpy as np
import matplotlib.pyplot as plt
from sklearn.datasets import make_classification
aa.options["verbose"] = False
aa.plot_settings()
X, labels = make_classification(n_samples=160, n_features=10, n_informative=6, random_state=42)
X_train, labels_train = X[:120], labels[:120]
# include one clearly out-of-distribution sample so the trust status is visible
X_new = np.vstack([X[120:], (X_train[labels_train == 1][0] + 20.0)[None, :]])
rm = aa.ReliabilityModel(random_state=42).fit(X=X_train, labels=labels_train)
df_rel = rm.predict(X=X_new)
df_eval = rm.eval(X=X[120:], labels=labels[120:])

aa.ReliabilityModelPlot().ranking(df_rel=df_rel,
                                  names=[f"sample {i}" for i in range(len(df_rel))],
                                  figsize=(5, 8), top_n=15, title="Per-sample reliability",
                                  ax=None)
plt.tight_layout()
plt.show()