Updated recon error histogram for multi alphas/bins

notebooks/demo/mnist_dataset.ipynb

@@ -196,7 +196,7 @@
     "mnist_recon_error = ReconstructionError(autoencoder, test_ds, axis=(1, 2))\n",
     "\n",
     "# view distribution\n",
-    "mnist_recon_error.view_error_distribution(\"MNIST Autoencoder: Reconstruction Error Distribution\")"
+    "mnist_recon_error.histogram(\"MNIST Autoencoder: Reconstruction Error Distribution\")"
    ]
   }
  ],

notebooks/demo/tf_flowers.ipynb

@@ -206,7 +206,7 @@
     "tfflower_recon_error = ReconstructionError(autoencoder, x_val)\n",
     "\n",
     "# view distribution\n",
-    "tfflower_recon_error.view_error_histogram(\"tf_flowers Autoencoder: Reconstruction Error Distribution\", bins=100)"
+    "tfflower_recon_error.histogram(\"tf_flowers Autoencoder: Reconstruction Error Distribution\", bins=[100])"
    ]
   }
  ],

src/autoencoder/data/anomaly.py

@@ -102,15 +102,16 @@ def gen_reconstruction_error(self) -> Generator[Any, None, None]:
             # update errors list
             yield from mse.numpy()
 
-    def view_error_histogram(
+    def _plot_error_distribution(
         self,
-        title: str = "Reconstruction Error Histogram",
-        bins: int = 10**3,
-        label: str = "threshold_source",
-        additional_data: Optional[List["ReconstructionError"]] = None,
-        additional_labels: Optional[List[str]] = None,
+        title: str,
+        label: str,
+        density: bool,
+        additional_data: Optional[List["ReconstructionError"]],
+        additional_labels: Optional[List[str]],
+        alphas: Optional[List[float]],
+        bins: Optional[List[int]],
     ) -> None:
-        """Plot the reconstruction error distribution."""
         # setup list of data and labels
         error_data = [self.errors["reconstruction_error"].values.tolist()]
 
@@ -130,7 +131,66 @@ def view_error_histogram(
             # otherwise don't use any
             error_labels = None
 
+        # check for alphas
+        if alphas is None:
+            # determine alpha value
+            alph_val = 0.5 if len(error_data) > 1 else 1
+
+            # build alphas list
+            alphas = [alph_val] * len(error_data)
+
+        # check for bins
+        if bins is None:
+            # build default bins list
+            bins = [10**3] * len(error_data)
+
         # now plot
         plot_error_distribution(
-            error_data, self.threshold, bins, title, labels=error_labels
+            errors=error_data,
+            threshold=self.threshold,
+            title=title,
+            bins=bins,
+            alphas=alphas,
+            labels=error_labels,
+            density=density,
+        )
+
+    def histogram(
+        self,
+        title: str = "Reconstruction Error Histogram",
+        label: str = "threshold_source",
+        additional_data: Optional[List["ReconstructionError"]] = None,
+        additional_labels: Optional[List[str]] = None,
+        alphas: Optional[List[float]] = None,
+        bins: Optional[List[int]] = None,
+    ) -> None:
+        """Plot the reconstruction error as a histogram."""
+        self._plot_error_distribution(
+            title=title,
+            bins=bins,
+            label=label,
+            additional_data=additional_data,
+            additional_labels=additional_labels,
+            density=False,
+            alphas=alphas,
+        )
+
+    def probability_distribution(
+        self,
+        title: str = "Reconstruction Error Probability Distribution",
+        label: str = "threshold_source",
+        additional_data: Optional[List["ReconstructionError"]] = None,
+        additional_labels: Optional[List[str]] = None,
+        alphas: Optional[List[float]] = None,
+        bins: Optional[List[int]] = None,
+    ) -> None:
+        """Plot the reconstruction error as a probability distribution."""
+        self._plot_error_distribution(
+            title=title,
+            bins=bins,
+            label=label,
+            additional_data=additional_data,
+            additional_labels=additional_labels,
+            density=True,
+            alphas=alphas,
         )

src/autoencoder/data/visualize.py

@@ -59,17 +59,20 @@ def compare_image_predictions(
 def plot_error_distribution(
     errors: List[List[float]],
     threshold: float,
-    bins: int,
     title: str,
+    bins: List[int],
+    alphas: List[float],
     density: bool = False,
     labels: Optional[List[str]] = None,
 ) -> None:
     """Plot a simple histogram for the reconstruction error."""
-    # calculate alpha
-    alpha = 0.5 if len(errors) > 1 else None
-
     # build histogram
-    plt.hist(x=errors, alpha=alpha, bins=bins, density=density)
+    for (
+        err_data,
+        alph,
+        bn,
+    ) in zip(errors, alphas, bins, strict=True):
+        plt.hist(x=err_data, alpha=alph, bins=bn, density=density, stacked=False)
 
     # add title
     plt.title(title)