import unittest
import importlib
import numpy as np
from sklearn.datasets import make_classification
from sklearn.ensemble import IsolationForest
from sklearn.decomposition import PCA
from sklearn.cluster import KMeans
from sklearn.metrics import (
    precision_score, recall_score, roc_auc_score,
    silhouette_score, davies_bouldin_score, calinski_harabasz_score
)

def _dynamic_test(test_case, condition, success_msg, failure_msg):
    if condition:
        test_case._testMethodName = success_msg
        test_case.assertTrue(True, success_msg)
    else:
        test_case._testMethodName = failure_msg
        test_case.fail(failure_msg)

class TestUnsupervisedMetrics(unittest.TestCase):
    @classmethod
    def setUpClass(cls):
        cls.user = importlib.import_module("user_code")

        X, y_true = make_classification(
            n_samples=1000, n_features=6, n_informative=4, n_redundant=0,
            weights=[0.95, 0.05], random_state=42
        )

        # Anomaly detection
        model = IsolationForest(contamination=0.05, random_state=42)
        y_pred = model.fit_predict(X)
        y_pred = np.where(y_pred == -1, 1, 0)

        cls.precision = precision_score(y_true, y_pred)
        cls.recall = recall_score(y_true, y_pred)
        cls.roc_auc = roc_auc_score(y_true, y_pred)

        # PCA
        pca = PCA(n_components=2, random_state=42)
        X_pca = pca.fit_transform(X)
        X_reconstructed = pca.inverse_transform(X_pca)

        cls.explained_variance = np.sum(pca.explained_variance_ratio_)
        cls.reconstruction_error = np.mean((X - X_reconstructed) ** 2)

        # KMeans
        kmeans = KMeans(n_clusters=3, random_state=42)
        labels = kmeans.fit_predict(X_pca)

        cls.inertia = kmeans.inertia_
        cls.silhouette = silhouette_score(X_pca, labels)
        cls.davies = davies_bouldin_score(X_pca, labels)
        cls.calinski = calinski_harabasz_score(X_pca, labels)

    def test_anomaly_precision(self):
        cond = abs(self.user.precision - self.precision) < 1e-6
        _dynamic_test(self, cond, "Anomaly precision correct.", "Anomaly precision incorrect.")

    def test_anomaly_recall(self):
        cond = abs(self.user.recall - self.recall) < 1e-6
        _dynamic_test(self, cond, "Anomaly recall correct.", "Anomaly recall incorrect.")

    def test_anomaly_roc_auc(self):
        cond = abs(self.user.roc_auc - self.roc_auc) < 1e-6
        _dynamic_test(self, cond, "Anomaly ROC–AUC correct.", "Anomaly ROC–AUC incorrect.")

    def test_pca_explained_variance(self):
        cond = abs(self.user.explained_variance - self.explained_variance) < 1e-6
        _dynamic_test(self, cond, "PCA explained variance correct.", "PCA explained variance incorrect.")

    def test_pca_reconstruction_error(self):
        cond = abs(self.user.reconstruction_error - self.reconstruction_error) < 1e-6
        _dynamic_test(self, cond, "PCA reconstruction error correct.", "PCA reconstruction error incorrect.")

    def test_clustering_metrics(self):
        cond = (
            abs(self.user.inertia - self.inertia) < 1e-6 and
            abs(self.user.silhouette - self.silhouette) < 1e-6 and
            abs(self.user.davies - self.davies) < 1e-6 and
            abs(self.user.calinski - self.calinski) < 1e-6
        )
        _dynamic_test(self, cond, "Clustering metrics computed correctly.", "Clustering metrics incorrect.")

if __name__ == "__main__":
    unittest.main(argv=["first-arg-is-ignored"], exit=False)

test_main.py

A comprehensive course for intermediate learners to master the key evaluation metrics used in machine learning, covering both supervised and unsupervised tasks. The course progresses from foundational concepts to practical Python implementations, interpretation of results, and a comparative overview of metric types and their trade-offs.

Explore the fundamental metrics used to evaluate classification models, including their definitions, formulas, Python implementations, and interpretation. Includes advanced metrics and model validation techniques relevant to classification.

Delve into the core metrics for evaluating regression models, including their mathematical foundations, Python implementations, and result interpretation. Includes advanced validation techniques relevant to regression.

Examine the key metrics for evaluating unsupervised learning models, including clustering, dimensionality reduction, and anomaly detection. Learn their mathematical foundations, Python implementations, and interpretation.

Challenge: Unsupervised Metrics

1. Anomaly Detection Evaluation

2. Dimensionality Reduction Evaluation

3. Clustering Evaluation

Ratkaisu

Awesome!

Challenge: Unsupervised Metrics

1. Anomaly Detection Evaluation

2. Dimensionality Reduction Evaluation

3. Clustering Evaluation

Ratkaisu