fix(fit-method): Change the range of fitting exceedances (#29)

Co-authored-by: N. L <nino@pleno.earth>

pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "setuptools.build_meta"
 name = "anomalytics"
 description = "The ultimate anomaly detection library."
 readme = "README.md"
-version = "0.1.4"
+version = "0.1.5"
 license = {file = "LICENSE"}
 requires-python = ">=3.10"
 authors = [

src/anomalytics/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.1.4"
+__version__ = "0.1.5"
 
 __all__ = [
     "get_anomaly",

src/anomalytics/models/peaks_over_threshold.py

@@ -550,7 +550,7 @@ def plot(
         alpha: float = 0.8,
     ):
         if isinstance(self.__exceedance, pd.Series):
-            nonzero_exceedences = [exceedence for exceedence in self.__exceedance.values if exceedence > 0]
+            nonzero_exceedences = self.__exceedance[self.__exceedance.values > 0]
         if plot_type == "l":
             plot_line(
                 dataset=self.__dataset,

src/anomalytics/stats/peaks_over_threshold.py

@@ -215,7 +215,7 @@ def get_anomaly_score(ts: pd.Series, t0: int, gpd_params: typing.Dict) -> pd.Ser
     t1_t2_exceedances = ts.iloc[t0:]
 
     for row in range(0, t1_t2_exceedances.shape[0]):
-        fit_exceedances = ts.iloc[t0 + row :]
+        fit_exceedances = ts.iloc[: t0 + row]
         future_exeedance = t1_t2_exceedances.iloc[row]
         nonzero_fit_exceedances = fit_exceedances[fit_exceedances.values > 0.0]
         if future_exeedance > 0:

tests/test_version.py

@@ -2,4 +2,4 @@
 
 
 def test_pkg_version():
-    assert __version__ == "0.1.4"
+    assert __version__ == "0.1.5"

tests/unit/detectors/test_pot_detector.py

@@ -58,18 +58,18 @@ def test_fit_with_genpareto_method(self):
         self.pot1_detector.fit()
 
         expected_anomaly_scores = pd.Series(
-            data=[1.922777880970598, 2.445890926224859, 3.6935717350888506, 3121651314.625431],
+            data=[float("inf"), float("inf"), float("inf"), float("inf")],
             index=self.sample_1_ts.index[6:],  # type: ignore
             name="anomaly scores",
         )
         expected_params = {
             0: {
                 "index": pd.Timestamp("2023-01-07 00:00:00"),
-                "c": -1.6804238287454643,
+                "c": -2.687778724221391,
                 "loc": 0,
-                "scale": 1.5123814458709186,
-                "p_value": 0.5200808735615424,
-                "anomaly_score": 1.922777880970598,
+                "scale": 1.3438893621106958,
+                "p_value": 0.0,
+                "anomaly_score": float("inf"),
             },
         }
 
@@ -78,7 +78,7 @@ def test_fit_with_genpareto_method(self):
 
     def test_compute_anomaly_threshold_method(self):
         expected_detected_data = [True]
-        expected_anomaly_threshold = 1.2394417670604552
+        expected_anomaly_threshold = 1.8927400332325932
 
         self.pot2_detector.get_extremes(q=0.90)
         self.pot2_detector.fit()
@@ -96,11 +96,11 @@ def test_evaluation_with_ks_1sample(self):
         expected_kstest_result = pd.DataFrame(
             data={
                 "total_nonzero_exceedances": [50],
-                "stats_distance": [0.8331661595029082],
-                "p_value": [4.312557311051241e-05],
-                "c": [-1.3371948412738648],
+                "stats_distance": [0.33333333326007464],
+                "p_value": [0.4234396436048128],
+                "c": [-1.5741853768217173],
                 "loc": [0],
-                "scale": [272179.457686573],
+                "scale": [71.62543464538814],
             }
         )