Bug fix: Assure that feature dataframe indices are unique

tobac/feature_detection.py

@@ -504,7 +504,9 @@ def feature_detection_multithreshold_timestep(
             idx_start=idx_start,
         )
         if any([x is not None for x in features_threshold_i]):
-            features_thresholds = pd.concat([features_thresholds, features_threshold_i])
+            features_thresholds = pd.concat(
+                [features_thresholds, features_threshold_i], ignore_index=True
+            )
 
         # For multiple threshold, and features found both in the current and previous step, remove "parent" features from Dataframe
         if i_threshold > 0 and not features_thresholds.empty and regions_old:

tobac/tests/test_feature_detection.py

@@ -11,10 +11,9 @@ def test_feature_detection_multithreshold_timestep(
     test_threshs, dxy, wavelength_filtering
 ):
     """
-    Tests ```tobac.feature_detection.feature_detection_multithreshold_timestep
+    Tests ```tobac.feature_detection.feature_detection_multithreshold_timestep```
     """
     import numpy as np
-    from tobac import feature_detection
 
     # start by building a simple dataset with a single feature and seeing
     # if we identify it
@@ -37,7 +36,7 @@ def test_feature_detection_multithreshold_timestep(
         amplitude=test_amp,
     )
     test_data_iris = tbtest.make_dataset_from_arr(test_data, data_type="iris")
-    fd_output = feature_detection.feature_detection_multithreshold_timestep(
+    fd_output = feat_detect.feature_detection_multithreshold_timestep(
         test_data_iris,
         0,
         threshold=test_threshs,
@@ -53,6 +52,71 @@ def test_feature_detection_multithreshold_timestep(
     assert fd_output.iloc[0]["hdim_2"] == pytest.approx(test_hdim_2_pt)
 
 
+@pytest.mark.parametrize(
+    "test_threshs, min_distance, dxy", [([1, 2, 3], 100000, 10000)]
+)
+def test_filter_min_distance(test_threshs, min_distance, dxy):
+    """
+    Tests ```tobac.feature_detection.filter_min_distance```
+    """
+    # start by building a simple dataset with two features close to each other
+    import numpy as np
+
+    test_dset_size = (50, 50)
+    test_hdim_1_pt = 20.0
+    test_hdim_2_pt = 20.0
+    test_hdim_1_sz = 5
+    test_hdim_2_sz = 5
+    test_amp = 5
+    test_min_num = 2
+
+    test_data = np.zeros(test_dset_size)
+    test_data = tbtest.make_feature_blob(
+        test_data,
+        test_hdim_1_pt,
+        test_hdim_2_pt,
+        h1_size=test_hdim_1_sz,
+        h2_size=test_hdim_2_sz,
+        amplitude=test_amp,
+    )
+
+    ## add another blob with smaller value
+    test_hdim_1_pt2 = 25.0
+    test_hdim_2_pt2 = 25.0
+    test_hdim_1_sz2 = 2
+    test_hdim_2_sz2 = 2
+    test_amp2 = 3
+    test_data = tbtest.make_feature_blob(
+        test_data,
+        test_hdim_1_pt2,
+        test_hdim_2_pt2,
+        h1_size=test_hdim_1_sz2,
+        h2_size=test_hdim_2_sz2,
+        amplitude=test_amp2,
+    )
+    test_data_iris = tbtest.make_dataset_from_arr(test_data, data_type="iris")
+
+    # identify these features
+    fd_output = feat_detect.feature_detection_multithreshold_timestep(
+        test_data_iris,
+        0,
+        threshold=test_threshs,
+        n_min_threshold=test_min_num,
+        min_distance=min_distance,
+        dxy=dxy,
+    )
+
+    # check if it function to filter
+    fd_filtered = feat_detect.filter_min_distance(fd_output, dxy, min_distance)
+
+    # Make sure we have only one feature (small feature in minimum distance should be removed )
+    assert len(fd_output.index) == 2
+    assert len(fd_filtered.index) == 1
+    # Make sure that the locations of the features is correct (should correspond to locations of first feature)
+    assert fd_filtered.iloc[0]["hdim_1"] == pytest.approx(test_hdim_1_pt)
+    assert fd_filtered.iloc[0]["hdim_2"] == pytest.approx(test_hdim_2_pt)
+
+
 @pytest.mark.parametrize(
     "position_threshold", [("center"), ("extreme"), ("weighted_diff"), ("weighted_abs")]
 )