Fixed Zscore implementation

.github/workflows/python-package.yml

@@ -39,7 +39,7 @@ jobs:
         flake8 . --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics
     - name: Test with pytest
       run: |
-        pytest -vs --doctest-modules --cov-fail-under=90 --cov-branch --cov=pythresh --cov-report term-missing --pyargs pythresh --continue-on-collection-errors -o log_cli=true -o log_level=DEBUG
+        pytest -vs --doctest-modules --cov-fail-under=90 --cov-branch --cov=pythresh --cov-report term-missing --pyargs pythresh --continue-on-collection-errors
     - name: Codecov
       uses: codecov/codecov-action@v4
       env:

docs/benchmark.rst

@@ -13,7 +13,7 @@ datasets may help when selecting based on either accuracy or robustness
 or both. PyOD provides a highly detailed analysis on the performance of
 all the available methods, with great insight and interpretability
 `anomaly detection benchmark paper
-<https://www.andrew.cmu.edu/user/yuezhao2/papers/22-neurips-adbench.pdf>`_.
+<https://proceedings.neurips.cc/paper_files/paper/2022/hash/cf93972b116ca5268827d575f2cc226b-Abstract-Datasets_and_Benchmarks.html>`_.
 
 Since the thresholding methods are dependant on both the dataset and the
 outlier detection likelihood scores, in order to quantify how well a

pythresh/test/test_zscore.py

@@ -48,7 +48,7 @@ def test_prediction_labels(self):
         for scores in self.all_scores:
 
             pred_labels = self.thres.eval(scores)
-            assert (self.thres.thresh_ is None)
+            assert (self.thres.thresh_ is not None)
             assert (self.thres.dscores_ is not None)
 
             assert (self.thres.dscores_.min() == 0)

pythresh/thresholds/zscore.py

@@ -72,13 +72,13 @@ def eval(self, decision):
         self.dscores_ = decision
 
         # Get the zscore of the decision scores
-        zscore = np.abs(stats.zscore(decision))
+        zscore = stats.zscore(decision)
 
         # Set the limit to where the zscore is 1
-        scores = np.zeros(len(decision), dtype=int)
+        labels = np.zeros(len(decision), dtype=int)
         mask = np.where(zscore >= 1.0)
-        scores[mask] = 1
+        labels[mask] = 1
 
-        self.thresh_ = None
+        self.thresh_ = np.min(labels[labels == 1])
 
-        return scores
+        return labels