From 9d318653c001287bcc8ae9d8e09d0187413cbed6 Mon Sep 17 00:00:00 2001
From: Ashley Xu <139821907+ashleyxuu@users.noreply.github.com>
Date: Tue, 27 Feb 2024 07:30:58 -0800
Subject: [PATCH] feat: Add ml.metrics.pairwise.manhattan_distance (#392)

---
 bigframes/ml/metrics/pairwise.py                 | 16 ++++++++++++++++
 tests/system/small/ml/test_metrics_pairwise.py   | 14 ++++++++++++++
 .../sklearn/metrics/pairwise.py                  | 15 +++++++++++++++
 3 files changed, 45 insertions(+)

diff --git a/bigframes/ml/metrics/pairwise.py b/bigframes/ml/metrics/pairwise.py
index 35b64c7850..9ebea4ef42 100644
--- a/bigframes/ml/metrics/pairwise.py
+++ b/bigframes/ml/metrics/pairwise.py
@@ -34,3 +34,19 @@ def paired_cosine_distances(
 paired_cosine_distances.__doc__ = inspect.getdoc(
     vendored_metrics_pairwise.paired_cosine_distances
 )
+
+
+def paired_manhattan_distance(
+    X: Union[bpd.DataFrame, bpd.Series], Y: Union[bpd.DataFrame, bpd.Series]
+) -> bpd.DataFrame:
+    X, Y = utils.convert_to_dataframe(X, Y)
+    if len(X.columns) != 1 or len(Y.columns) != 1:
+        raise ValueError("Inputs X and Y can only contain 1 column.")
+
+    base_bqml = core.BaseBqml(session=X._session)
+    return base_bqml.distance(X, Y, type="MANHATTAN", name="manhattan_distance")
+
+
+paired_manhattan_distance.__doc__ = inspect.getdoc(
+    vendored_metrics_pairwise.paired_manhattan_distance
+)
diff --git a/tests/system/small/ml/test_metrics_pairwise.py b/tests/system/small/ml/test_metrics_pairwise.py
index 47bd1e18d0..e2aee971ee 100644
--- a/tests/system/small/ml/test_metrics_pairwise.py
+++ b/tests/system/small/ml/test_metrics_pairwise.py
@@ -33,3 +33,17 @@ def test_paired_cosine_distances():
     pd.testing.assert_frame_equal(
         result.to_pandas(), expected_pd_df, check_dtype=False, check_index_type=False
     )
+
+
+def test_paired_manhattan_distance():
+    x_col = [np.array([4.1, 0.5, 1.0])]
+    y_col = [np.array([3.0, 0.0, 2.5])]
+    X = bpd.read_pandas(pd.DataFrame({"X": x_col}))
+    Y = bpd.read_pandas(pd.DataFrame({"Y": y_col}))
+
+    result = metrics.pairwise.paired_manhattan_distance(X, Y)
+    expected_pd_df = pd.DataFrame({"X": x_col, "Y": y_col, "manhattan_distance": [3.1]})
+
+    pd.testing.assert_frame_equal(
+        result.to_pandas(), expected_pd_df, check_dtype=False, check_index_type=False
+    )
diff --git a/third_party/bigframes_vendored/sklearn/metrics/pairwise.py b/third_party/bigframes_vendored/sklearn/metrics/pairwise.py
index c309b08d88..5791d850ff 100644
--- a/third_party/bigframes_vendored/sklearn/metrics/pairwise.py
+++ b/third_party/bigframes_vendored/sklearn/metrics/pairwise.py
@@ -24,3 +24,18 @@ def paired_cosine_distances(X, Y) -> bpd.DataFrame:
         bigframes.dataframe.DataFrame: DataFrame with columns of X, Y and cosine_distance
     """
     raise NotImplementedError(constants.ABSTRACT_METHOD_ERROR_MESSAGE)
+
+
+def paired_manhattan_distance(X, Y) -> bpd.DataFrame:
+    """Compute the L1 distances between the vectors in X and Y.
+
+    Args:
+        X (Series or single column DataFrame of array of numeric type):
+            Input data.
+        Y (Series or single column DataFrame of array of numeric type):
+            Input data. X and Y are mapped by indexes, must have the same index.
+
+    Returns:
+        bigframes.dataframe.DataFrame: DataFrame with columns of X, Y and manhattan_distance
+    """
+    raise NotImplementedError(constants.ABSTRACT_METHOD_ERROR_MESSAGE)