Bug: Power Module `centroid_diffs`

[ssolson]

@tnelson-integral I believe there is a bug in centroid_diffs calculation.

https://github.com/sandialabs/seat-qgis-plugin/blob/d04f83e5584b713f5e9ba728759a9a1dd54461b0/seat/power_module.py#L125-L147

I created the following test using the euclidean distance to calculate the distance

    def test_centroid_diffs(self):
        """
        Test the centroid_diffs function, considering the first value in each centroid as an index.
        """
        # Prepare test inputs
        test_centroids = np.array([
            [0, 2.0, 2.0],  # Index 0, Coordinates (2.0, 2.0)
            [1, 5.0, 5.0],  # Index 1, Coordinates (5.0, 5.0)
            [2, 7.0, 8.0]   # Index 2, Coordinates (7.0, 8.0)
        ])
        test_centroid = np.array([3, 6.0, 6.0])  # Index 3, Coordinates (6.0, 6.0)

        # Expected output: Closest centroid to test_centroid
        def calculate_expected_pair(centroids, centroid):
            # Calculate Euclidean distances, excluding the index
            distances = np.sqrt(np.sum((centroids[:, 1:] - centroid[1:])**2, axis=1))

            # Find the index of the centroid with the minimum distance
            closest_index = np.argmin(distances)

            # Return the expected pair: [index of test_centroid, index of closest centroid in test_centroids]
            return [int(centroid[0]), int(centroids[closest_index, 0])]

        expected_pair = self.calculate_expected_pair(test_centroids, test_centroid)

        # Call the function
        result_pair = pm.centroid_diffs(test_centroids, test_centroid)

        # Assert the result
        self.assertEqual(result_pair, expected_pair, "The identified closest centroid pair is incorrect.")

The expect pair differs: AssertionError: Lists differ: [3, 0] != [3, 1]

Am I missing something about the data and the way it should be handled?

[ssolson]

I think this would be the correct implementation:

def centroid_diffs(Centroids, centroid):
    # Calculate Euclidean distances
    distances = np.sqrt(np.sum((Centroids[:, 1:] - centroid[1:])**2, axis=1))

    # Find the index of the centroid with the minimum distance
    min_arg = np.argmin(distances)

    # Return the pair: index of the single centroid and the index of the closest centroid
    pair = [int(centroid[0]), int(Centroids[min_arg, 0])]
    return pair

[tnelson-integral]

While the current method works for the tutorial dataset (see image), it is not robust for different configurations. The suggested method is more robust and fixes this issue.