fix(gpd-plot): handle non-zero params and exceedances in POTDetector (

#26) * fix(gpd_plot): add params processing before using as argument for "gpd+ov" * fix(gpd_plot): handle non-zero exceedances in models, not in plot * refactor(example-notebook): create new flow for the notebook using `POTDetector` * fix(pot-detector-plot): handle non-zero anomaly scores in method instead in plot function --------- Co-authored-by: N. L <nino@pleno.earth>
Aeternalis-Ingenium · Dec 11, 2023 · b4aa501 · b4aa501
1 parent feaffa3
commit b4aa501
Show file tree

Hide file tree

Showing 3 changed files with 6,182 additions and 302 deletions.
diff --git a/docs/example.ipynb b/docs/example.ipynb
diff --git a/src/anomalytics/models/peaks_over_threshold.py b/src/anomalytics/models/peaks_over_threshold.py
@@ -549,6 +549,8 @@ def plot(
         th_line_width: int = 2,
         alpha: float = 0.8,
     ):
+        if isinstance(self.__exceedance, pd.Series):
+            nonzero_exceedences = [exceedence for exceedence in self.__exceedance.values if exceedence > 0]
         if plot_type == "l":
             plot_line(
                 dataset=self.__dataset,
@@ -566,8 +568,10 @@ def plot(
                 alpha=alpha,
             )
         elif plot_type == "l+ath":
+            if isinstance(self.__anomaly_score, pd.Series):
+                nonzero_anomaly_scores = self.__anomaly_score[self.__anomaly_score.values > 0]
             plot_line(
-                dataset=self.__exceedance,
+                dataset=nonzero_anomaly_scores,
                 threshold=self.__anomaly_threshold,
                 title=title,
                 xlabel=xlabel,
@@ -611,7 +615,7 @@ def plot(
             )
         elif plot_type == "gpd":
             plot_gen_pareto(
-                dataset=self.__exceedance,
+                dataset=nonzero_exceedences,
                 title=title,
                 xlabel=xlabel,
                 ylabel=ylabel,
@@ -623,8 +627,9 @@ def plot(
                 params=None,
             )
         elif plot_type == "gpd+ov":
+            last_nonzero_params = self.__get_nonzero_params[-1]
             plot_gen_pareto(
-                dataset=self.__exceedance,
+                dataset=nonzero_exceedences,
                 title=title,
                 xlabel=xlabel,
                 ylabel=ylabel,
@@ -633,7 +638,7 @@ def plot(
                 plot_height=plot_height,
                 plot_color=plot_color,
                 alpha=alpha,
-                params=self.__params,
+                params=last_nonzero_params,
             )
 
     def __str__(self) -> str:

diff --git a/src/anomalytics/plots/plot.py b/src/anomalytics/plots/plot.py
@@ -79,17 +79,16 @@ def plot_gen_pareto(
     plot_height: int = 8,
     plot_color: str = "black",
     alpha: float = 0.8,
-    params: typing.Union[typing.Dict, None] = None,
+    params: typing.Optional[typing.Dict] = None,
 ):
     fig = plt.figure(figsize=(plot_width, plot_height))
 
-    nonzero_exceedences = [exceedence for exceedence in dataset if exceedence > 0]
     if params:
         param_label = f"\n{round(params['c'], 3)}\n{round(params['loc'], 3)}\n{round(params['scale'], 3)}\n"
         overlay = np.linspace(
             stats.genpareto.ppf(0.1, c=params["c"], loc=params["loc"], scale=params["scale"]),
             stats.genpareto.ppf(0.999, c=params["c"], loc=params["loc"], scale=params["scale"]),
-            len(nonzero_exceedences),
+            len(dataset),
         )
         plt.plot(
             overlay,
@@ -99,12 +98,12 @@ def plot_gen_pareto(
             label=f"\nFitted Params:{param_label}",
         )
     plt.hist(
-        nonzero_exceedences,
+        dataset,
         bins=bins,
         density=True,
         alpha=alpha,
         color=plot_color,
-        label=f"{len(nonzero_exceedences)}",
+        label=f"{len(dataset)}",
     )
     plt.xlabel(xlabel)
     plt.ylabel(ylabel)