Add recall_score function in Ivy with Test

ivy/functional/frontends/sklearn/metrics/_classification.py

@@ -1,6 +1,7 @@
 import ivy
 from ivy.functional.frontends.numpy.func_wrapper import to_ivy_arrays_and_back
 from sklearn.utils.multiclass import type_of_target
+from ivy.utils.exceptions import IvyValueError
 
 
 @to_ivy_arrays_and_back
@@ -17,3 +18,34 @@ def accuracy_score(y_true, y_pred, *, normalize=True, sample_weight=None):
         ret = ret / y_true.shape[0]
         ret = ret.astype("float64")
     return ret
+
+
+@to_ivy_arrays_and_back
+def recall_score(y_true, y_pred, *, average="binary", sample_weight=None):
+    # Determine the type of the target variable
+    y_type = type_of_target(y_true)
+    # Check if the 'average' parameter has a valid value
+    if average != "binary" and average != "micro" and average != "macro":
+        raise IvyValueError(
+            "Invalid value for 'average'. Supported values are 'binary', 'micro', or"
+            " 'macro'."
+        )
+    # Calculate true positive and actual positive counts based on the target type
+    if y_type.startswith("multilabel"):
+        true_positive = ivy.sum(ivy.logical_and(y_true, y_pred) * sample_weight, axis=1)
+        actual_positive = ivy.sum(y_true * sample_weight, axis=1)
+    else:
+        true_positive = ivy.sum(ivy.logical_and(y_true, y_pred) * sample_weight)
+        actual_positive = ivy.sum(y_true * sample_weight)
+    # Calculate recall for each class or overall
+    recall = true_positive / ivy.maximum(
+        actual_positive, ivy.to_scalar(ivy.array([1], "float64"))
+    )
+    # Perform additional calculations for micro or macro averaging
+    if average == "micro":
+        recall = ivy.sum(true_positive) / ivy.maximum(
+            ivy.sum(actual_positive), ivy.to_scalar(ivy.array([1], "float64"))
+        )
+    elif average == "macro":
+        recall = ivy.mean(recall)
+    return recall

ivy_tests/test_ivy/test_frontends/test_sklearn/test_metrics/test_classification.py

@@ -5,6 +5,49 @@
 import numpy as np
 
 
+@handle_frontend_test(
+    fn_tree="your.module.path.recall_score",  # Update with the correct module path
+    arrays_and_dtypes=helpers.dtype_and_values(
+        available_dtypes=helpers.get_dtypes("float_and_integer"),
+        num_arrays=2,
+        min_value=-2,
+        max_value=2,
+        shared_dtype=True,
+        shape=(helpers.ints(min_value=2, max_value=5)),
+    ),
+    average=st.sampled_from(["binary", "micro", "macro"]),
+    sample_weight=st.none() | st.floats(min_value=0, max_value=1),
+)
+def test_recall_score(
+    arrays_and_dtypes,
+    on_device,
+    fn_tree,
+    frontend,
+    test_flags,
+    backend_fw,
+    average,
+    sample_weight,
+):
+    dtypes, values = arrays_and_dtypes
+    # Ensure discrete values if float dtype
+    for i in range(2):
+        if "float" in dtypes[i]:
+            values[i] = np.floor(values[i])
+
+    helpers.test_frontend_function(
+        input_dtypes=dtypes,
+        backend_to_test=backend_fw,
+        test_flags=test_flags,
+        fn_tree=fn_tree,
+        frontend=frontend,
+        on_device=on_device,
+        y_true=values[0],
+        y_pred=values[1],
+        average=average,
+        sample_weight=sample_weight,
+    )
+
+
 @handle_frontend_test(
     fn_tree="sklearn.metrics.accuracy_score",
     arrays_and_dtypes=helpers.dtype_and_values(