FERL Example

This example trains a compact FERL classifier and inspects both point and native evidential predictions.

import numpy as np
from ex_fuzzy import FERL
from sklearn.datasets import load_iris
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split

iris = load_iris(as_frame=True)
X_train, X_test, y_train, y_test = train_test_split(
    iris.data,
    iris.target,
    test_size=0.25,
    random_state=0,
    stratify=iris.target,
)

ferl = FERL(
    max_rules=10,
    max_depth=5,
    min_improvement=0.0,
    random_state=0,
)
ferl.fit(X_train, y_train, patience=5)

predictions = ferl.predict(X_test)
betp, belief, plausibility, ignorance = ferl.predict_credal(X_test)
prediction_sets = ferl.predict_set(X_test)

print(f"accuracy={accuracy_score(y_test, predictions):.3f}")
print(f"rules={ferl.n_rules()}")
print(f"mean ignorance={ignorance.mean():.3f}")
print(f"mean set size={prediction_sets.sum(axis=1).mean():.2f}")

for row in range(5):
    labels = iris.target_names[ferl.classes_[prediction_sets[row]]]
    print(
        row,
        labels.tolist(),
        np.round(betp[row], 3),
        round(float(ignorance[row]), 3),
    )

ferl.print_tree()

The equivalent runnable script is Demos/demos_module/ferl_demo.py in the source distribution.