tests.py

from unittest import TestCase

from precision_recall_fscore_accuracy import PrecisionRecallFscoreAccuracy


class PrecisionRecallFscoreAccuracyTestBinary(TestCase):
    """
    All of these tests are also used for multiset configuration. So please mind this and write the test in a way that
    it is valid for both configurations (do not use same label multiple times).
    """

    def setUp(self):
        self.metric = PrecisionRecallFscoreAccuracy(average="binary")

    def test_eok(self):
        self.assertDictEqual(
            {
                "precision": 1.0,
                "recall": 1.0,
                "fscore": 1.0,
                "accuracy": 1.0,
            },
            self.metric.compute(
                predictions=[0, 1, 0],
                references=[0, 1, 0]
            )
        )

    def test_eok_string(self):
        self.assertDictEqual(
            {
                "precision": 1.0,
                "recall": 1.0,
                "accuracy": 1.0,
                "fscore": 1.0
            },
            self.metric.compute(
                predictions=["0", "1", "0"],
                references=["0", "1", "0"]
            )
        )

    def test_completely_different(self):
        self.assertDictEqual(
            {
                "precision": 0.0,
                "recall": 0.0,
                "accuracy": 0.0,
                "fscore": 0.0
            },
            self.metric.compute(
                predictions=[0, 1, 0],
                references=[1, 0, 1]
            )
        )

    def test_max_precision(self):
        self.assertDictEqual(
            {
                "precision": 1.0,
                "recall": 0.5,
                "accuracy": 0.5,
                "fscore": 2 / 3
            },
            self.metric.compute(
                predictions=[0, 1],
                references=[1, 1]
            )
        )

    def test_max_recall(self):
        self.assertDictEqual(
            {
                "precision": 0.5,
                "recall": 1.0,
                "accuracy": 0.5,
                "fscore": 2 / 3
            },
            self.metric.compute(
                predictions=[1, 1],
                references=[1, 0]
            )
        )

    def test_partial_match(self):
        self.assertDictEqual(
            {
                "precision": 0.5,
                "recall": 0.5,
                "accuracy": 0.5,
                "fscore": 0.5
            },
            self.metric.compute(
                predictions=[0, 1, 0, 1],
                references=[1, 1, 0, 0]
            )
        )