From 702899ea0aeb57b5761b85c43d1416ca81d4da00 Mon Sep 17 00:00:00 2001
From: Jonathan de Bruin <jonathandebruinos@gmail.com>
Date: Thu, 14 Mar 2019 20:59:51 +0100
Subject: [PATCH] resolve conflict with threshold and missing value (#85)

Closes #70
---
 recordlinkage/compare.py |  9 +++++----
 tests/test_compare.py    | 42 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 47 insertions(+), 4 deletions(-)

diff --git a/recordlinkage/compare.py b/recordlinkage/compare.py
index b5ea9479..1a3153fc 100644
--- a/recordlinkage/compare.py
+++ b/recordlinkage/compare.py
@@ -151,13 +151,14 @@ def _compute_vectorized(self, s_left, s_right):
                 self.method))
 
         c = str_sim_alg(s_left, s_right)
-        c = _fillna(c, self.missing_value)
 
         if self.threshold is not None:
-            return (c >= self.threshold).astype(numpy.float64)
-        else:
-            return c
+            c = c.where((c < self.threshold) | (pandas.isnull(c)), other=1.0)
+            c = c.where((c >= self.threshold) | (pandas.isnull(c)), other=0.0)
 
+        c = _fillna(c, self.missing_value)
+
+        return c
 
 class Numeric(BaseCompareFeature):
     """Compute the (partial) similarity between numeric values.
diff --git a/tests/test_compare.py b/tests/test_compare.py
index a483a11a..c2eb10c7 100644
--- a/tests/test_compare.py
+++ b/tests/test_compare.py
@@ -1183,6 +1183,48 @@ def test_fuzzy(self):
         assert (result[result.notnull()] >= 0).all(1).all(0)
         assert (result[result.notnull()] <= 1).all(1).all(0)
 
+    def test_threshold(self):
+
+        A = DataFrame({'col': [u"gretzky", u"gretzky99", u"gretzky", u"gretzky"]})
+        B = DataFrame({'col': [u"gretzky", u"gretzky", nan, u"wayne"]})
+        ix = MultiIndex.from_arrays([A.index.values, B.index.values])
+
+        comp = recordlinkage.Compare()
+        comp.string(
+            'col',
+            'col',
+            method="levenshtein",
+            threshold=0.5,
+            missing_value=2.0,
+            label="x_col1"
+        )
+        comp.string(
+            'col',
+            'col',
+            method="levenshtein",
+            threshold=1.0,
+            missing_value=0.5,
+            label="x_col2"
+        )
+        comp.string(
+            'col',
+            'col',
+            method="levenshtein",
+            threshold=0.0,
+            missing_value=nan,
+            label="x_col3"
+        )
+        result = comp.compute(ix, A, B)
+
+        expected = Series([1.0, 1.0, 2.0, 0.0], index=ix, name="x_col1")
+        pdt.assert_series_equal(result["x_col1"], expected)
+
+        expected = Series([1.0, 0.0, 0.5, 0.0], index=ix, name="x_col2")
+        pdt.assert_series_equal(result["x_col2"], expected)
+
+        expected = Series([1.0, 1.0, nan, 1.0], index=ix, name="x_col3")
+        pdt.assert_series_equal(result["x_col3"], expected)
+
     @pytest.mark.parametrize("alg", STRING_SIM_ALGORITHMS)
     def test_incorrect_input(self, alg):