dataeval 0.86.3__py3-none-any.whl → 0.86.4__py3-none-any.whl

dataeval/__init__.py

@@ -8,7 +8,7 @@ shifts that impact performance of deployed models.
 from __future__ import annotations
 
 __all__ = ["config", "detectors", "log", "metrics", "typing", "utils", "workflows"]
-__version__ = "0.86.3"
+__version__ = "0.86.4"
 
 import logging
 

dataeval/data/_images.py

@@ -4,6 +4,8 @@ __all__ = []
 
 from typing import TYPE_CHECKING, Any, Generic, Iterator, Sequence, TypeVar, cast, overload
 
+import numpy as np
+
 from dataeval.typing import Array, ArrayLike, Dataset
 from dataeval.utils._array import as_numpy, channels_first_to_last
 
@@ -58,7 +60,7 @@ class Images(Generic[T]):
         num_images = len(indices)
         num_rows = (num_images + images_per_row - 1) // images_per_row
         fig, axes = plt.subplots(num_rows, images_per_row, figsize=figsize)
-        for i, ax in enumerate(axes.flatten()):
+        for i, ax in enumerate(np.asarray(axes).flatten()):
             image = channels_first_to_last(as_numpy(self[i]))
             ax.imshow(image)
             ax.axis("off")

dataeval/data/_metadata.py

@@ -20,6 +20,10 @@ from dataeval.utils._bin import bin_data, digitize_data
 from dataeval.utils.data.metadata import merge
 
 
+def _binned(name: str) -> str:
+    return f"{name}[]"
+
+
 @dataclass
 class FactorInfo:
     factor_type: Literal["categorical", "continuous", "discrete"] | None = None
@@ -157,13 +161,13 @@ class Metadata:
     def factor_names(self) -> Sequence[str]:
         """Factor names of the metadata."""
         self._structure()
-        return list(self._factors)
+        return list(filter(self._filter, self._factors))
 
     @property
     def factor_info(self) -> Mapping[str, FactorInfo]:
         """Factor types of the metadata."""
         self._bin()
-        return self._factors
+        return dict(filter(self._filter, self._factors.items()))
 
     @property
     def factor_data(self) -> NDArray[Any]:
@@ -195,13 +199,17 @@ class Metadata:
     @property
     def image_count(self) -> int:
         self._bin()
-        return int(self._image_indices.max() + 1)
+        return 0 if self._image_indices.size == 0 else int(self._image_indices.max() + 1)
+
+    def _filter(self, factor: str | tuple[str, Any]) -> bool:
+        factor = factor[0] if isinstance(factor, tuple) else factor
+        return factor in self.include if self.include else factor not in self.exclude
 
     def _reset_bins(self, cols: Iterable[str] | None = None) -> None:
         if self._is_binned:
             columns = self._dataframe.columns
-            for col in (col for col in cols or columns if f"{col}[|]" in columns):
-                self._dataframe.drop_in_place(f"{col}[|]")
+            for col in (col for col in cols or columns if _binned(col) in columns):
+                self._dataframe.drop_in_place(_binned(col))
                 self._factors[col] = FactorInfo()
             self._is_binned = False
 
@@ -244,7 +252,7 @@ class Metadata:
         bboxes = as_numpy(bboxes).astype(np.float32) if is_od else None
         srcidx = as_numpy(srcidx).astype(np.intp) if is_od else None
 
-        index2label = self._dataset.metadata.get("index2label", {})
+        index2label = self._dataset.metadata.get("index2label", {i: str(i) for i in np.unique(labels)})
 
         targets_per_image = None if srcidx is None else np.unique(srcidx, return_counts=True)[1].tolist()
         merged = merge(raw, return_dropped=True, ignore_lists=False, targets_per_image=targets_per_image)
@@ -260,8 +268,9 @@ class Metadata:
         }
 
         self._raw = raw
+        self._index2label = index2label
         self._class_labels = labels
-        self._class_names = [index2label.get(i, str(i)) for i in np.unique(labels)]
+        self._class_names = list(index2label.values())
         self._image_indices = target_dict["image_index"]
         self._factors = dict.fromkeys(factor_dict, FactorInfo())
         self._dataframe = pl.DataFrame({**target_dict, **factor_dict})
@@ -289,10 +298,10 @@ class Metadata:
             )
 
         column_set = set(df.columns)
-        for col in (col for col in self.factor_names if f"{col}[|]" not in column_set):
+        for col in (col for col in self.factor_names if _binned(col) not in column_set):
             # Get data as numpy array for processing
             data = df[col].to_numpy()
-            col_dz = f"{col}[|]"
+            col_dz = _binned(col)
             if col in factor_bins:
                 # User provided binning
                 bins = factor_bins[col]
@@ -326,23 +335,6 @@ class Metadata:
         self._factors.update(factor_info)
         self._is_binned = True
 
-    def get_factors_by_type(self, factor_type: Literal["categorical", "continuous", "discrete"]) -> Sequence[str]:
-        """
-        Get the names of factors of a specific type.
-
-        Parameters
-        ----------
-        factor_type : Literal["categorical", "continuous", "discrete"]
-            The type of factors to retrieve.
-
-        Returns
-        -------
-        list[str]
-            List of factor names of the specified type.
-        """
-        self._bin()
-        return [name for name, info in self.factor_info.items() if info.factor_type == factor_type]
-
     def add_factors(self, factors: Mapping[str, Array | Sequence[Any]]) -> None:
         """
         Add additional factors to the metadata.

dataeval/metadata/_distance.py

@@ -81,7 +81,7 @@ def metadata_distance(metadata1: Metadata, metadata2: Metadata) -> MetadataDista
     """
 
     _compare_keys(metadata1.factor_names, metadata2.factor_names)
-    cont_fnames = metadata1.get_factors_by_type("continuous")
+    cont_fnames = [name for name, info in metadata1.factor_info.items() if info.factor_type == "continuous"]
 
     if not cont_fnames:
         return MetadataDistanceOutput({})

dataeval/metrics/bias/_balance.py

@@ -99,9 +99,10 @@ def balance(
     factor_types = {"class_label": "categorical"} | {k: v.factor_type for k, v in metadata.factor_info.items()}
     is_discrete = [factor_type != "continuous" for factor_type in factor_types.values()]
     num_factors = len(factor_types)
+    class_labels = metadata.class_labels
 
     mi = np.full((num_factors, num_factors), np.nan, dtype=np.float32)
-    data = np.hstack((metadata.class_labels[:, np.newaxis], data))
+    data = np.hstack((class_labels[:, np.newaxis], data))
 
     for idx, factor_type in enumerate(factor_types.values()):
         if factor_type != "continuous":
@@ -132,12 +133,12 @@ def balance(
     factors = nmi[1:, 1:]
 
     # assume class is a factor
-    num_classes = len(metadata.class_names)
+    u_classes = np.unique(class_labels)
+    num_classes = len(u_classes)
     classwise_mi = np.full((num_classes, num_factors), np.nan, dtype=np.float32)
 
     # classwise targets
-    classes = np.unique(metadata.class_labels)
-    tgt_bin = data[:, 0][:, None] == classes
+    tgt_bin = data[:, 0][:, None] == u_classes
 
     # classification MI for discrete/categorical features
     for idx in range(num_classes):

dataeval/metrics/stats/_labelstats.py

@@ -2,7 +2,7 @@ from __future__ import annotations
 
 __all__ = []
 
-from typing import Any, Mapping, TypeVar
+from typing import Any, TypeVar
 
 import polars as pl
 
@@ -14,10 +14,6 @@ from dataeval.typing import AnnotatedDataset
 TValue = TypeVar("TValue")
 
 
-def _sort_to_list(d: Mapping[int, TValue]) -> list[TValue]:
-    return [t[1] for t in sorted(d.items())]
-
-
 @set_metadata
 def labelstats(dataset: Metadata | AnnotatedDataset[Any]) -> LabelStatsOutput:
     """
@@ -58,20 +54,20 @@ def labelstats(dataset: Metadata | AnnotatedDataset[Any]) -> LabelStatsOutput:
 
     # Count occurrences of each label across all images
     label_counts_df = metadata_df.group_by("class_label").len()
-    label_counts = label_counts_df.sort("class_label")["len"].to_list()
+    label_counts = dict(zip(label_counts_df["class_label"], label_counts_df["len"]))
 
     # Count unique images per label (how many images contain each label)
     image_counts_df = metadata_df.select(["image_index", "class_label"]).unique().group_by("class_label").len()
-    image_counts = image_counts_df.sort("class_label")["len"].to_list()
+    image_counts = dict(zip(image_counts_df["class_label"], image_counts_df["len"]))
 
     # Create index_location mapping (which images contain each label)
-    index_location: list[list[int]] = [[] for _ in range(len(metadata.class_names))]
+    index_location: dict[int, list[int]] = {}
     for row in metadata_df.group_by("class_label").agg(pl.col("image_index")).to_dicts():
         indices = row["image_index"]
         index_location[row["class_label"]] = sorted(dict.fromkeys(indices)) if isinstance(indices, list) else [indices]
 
     # Count labels per image
-    label_per_image_df = metadata_df.group_by("image_index").agg(pl.count().alias("label_count"))
+    label_per_image_df = metadata_df.group_by("image_index").agg(pl.len().alias("label_count"))
     label_per_image = label_per_image_df.sort("image_index")["label_count"].to_list()
 
     return LabelStatsOutput(
@@ -81,6 +77,6 @@ def labelstats(dataset: Metadata | AnnotatedDataset[Any]) -> LabelStatsOutput:
         image_indices_per_class=index_location,
         image_count=len(label_per_image),
         class_count=len(metadata.class_names),
-        label_count=sum(label_counts),
+        label_count=sum(label_counts.values()),
         class_names=metadata.class_names,
     )

dataeval/outputs/_linters.py

@@ -54,7 +54,7 @@ def _reorganize_by_class_and_metric(
     for img, group in result.items():
         for extreme in group:
             metrics.setdefault(extreme, []).append(img)
-            for i, images in enumerate(lstats.image_indices_per_class):
+            for i, images in lstats.image_indices_per_class.items():
                 if img in images:
                     class_wise[lstats.class_names[i]][extreme] = class_wise[lstats.class_names[i]].get(extreme, 0) + 1
 

dataeval/outputs/_stats.py

@@ -272,7 +272,7 @@ class LabelStatsOutput(Output):
     image_counts_per_class : Mapping[int, int]
         Dictionary whose keys are the different label classes and
         values are total counts of each image the class is present in
-    image_indices_per_class : Mapping[int, list]
+    image_indices_per_class : Mapping[int, Sequence[int]]
         Dictionary whose keys are the different label classes and
         values are lists containing the images that have that label
     image_count : int
@@ -284,10 +284,10 @@ class LabelStatsOutput(Output):
     class_names : Sequence[str]
     """
 
-    label_counts_per_class: Sequence[int]
+    label_counts_per_class: Mapping[int, int]
     label_counts_per_image: Sequence[int]
-    image_counts_per_class: Sequence[int]
-    image_indices_per_class: Sequence[Sequence[int]]
+    image_counts_per_class: Mapping[int, int]
+    image_indices_per_class: Mapping[int, Sequence[int]]
     image_count: int
     class_count: int
     label_count: int

dataeval/utils/_plot.py

@@ -164,9 +164,9 @@ def histogram_plot(
     rows = math.ceil(num_metrics / 3)
     cols = min(num_metrics, 3)
     fig, axs = plt.subplots(rows, 3, figsize=(cols * 3 + 1, rows * 3))
-
+    axs_flat = np.asarray(axs).flatten()
     for ax, metric in zip(
-        axs.flat,
+        axs_flat,
         data_dict,
     ):
         # Plot the histogram for the chosen metric
@@ -177,7 +177,7 @@ def histogram_plot(
         ax.set_ylabel(ylabel)
         ax.set_xlabel(xlabel)
 
-    for ax in axs.flat[num_metrics:]:
+    for ax in axs_flat[num_metrics:]:
         ax.axis("off")
         ax.set_visible(False)
 
@@ -222,9 +222,9 @@ def channel_histogram_plot(
     rows = math.ceil(num_metrics / 3)
     cols = min(num_metrics, 3)
     fig, axs = plt.subplots(rows, 3, figsize=(cols * 3 + 1, rows * 3))
-
+    axs_flat = np.asarray(axs).flatten()
     for ax, metric in zip(
-        axs.flat,
+        axs_flat,
         data_keys,
     ):
         # Plot the histogram for the chosen metric
@@ -246,7 +246,7 @@ def channel_histogram_plot(
         ax.set_ylabel(ylabel)
         ax.set_xlabel(xlabel)
 
-    for ax in axs.flat[num_metrics:]:
+    for ax in axs_flat[num_metrics:]:
         ax.axis("off")
         ax.set_visible(False)
 

{dataeval-0.86.3.dist-info → dataeval-0.86.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dataeval
-Version: 0.86.3
+Version: 0.86.4
 Summary: DataEval provides a simple interface to characterize image data and its impact on model performance across classification and object-detection tasks
 Home-page: https://dataeval.ai/
 License: MIT

{dataeval-0.86.3.dist-info → dataeval-0.86.4.dist-info}/RECORD

@@ -1,10 +1,10 @@
-dataeval/__init__.py,sha256=Z_VUOb2gf--uAtqeXyzIPUm11noNeEj16OSfkc6H6-Y,1636
+dataeval/__init__.py,sha256=6gfYCGo82QKKO58jQSma27Mr-R316vmCDbTjXRh5B7o,1636
 dataeval/_log.py,sha256=C7AGkIRzymvYJ0LQXtnShiy3i5Xrp8T58JzIHHguk_Q,365
 dataeval/config.py,sha256=hjad0TK1UmaKQlUuxqxt64_OAUqZkHjicBf06cvTyrQ,4082
 dataeval/data/__init__.py,sha256=wzQ6uUFLNB3VJR0a2QnRBYwEmwXT93q0WpHu7FmFW1E,486
 dataeval/data/_embeddings.py,sha256=PFjpdV9bfusCB4taTIYSzx1hP8nJb_KCkZTN8kMw-Hs,12885
-dataeval/data/_images.py,sha256=3d4Cv-xg5z6_LVtw1eL_QdFwzbDI1cwvPNQblkrMEMk,2622
-dataeval/data/_metadata.py,sha256=jEfGZhhvry7qtjU47VifL8ZO1hqXg1jntR3CztkaoWU,14462
+dataeval/data/_images.py,sha256=Rc_59CuU4zfN7Xm7an1XUx8ZghQg6a56VJWMZD9edRw,2654
+dataeval/data/_metadata.py,sha256=5pND6IZ5KeEGrhCDiBVxhU_BXWU0okBxt8oNkZ9a2_M,14309
 dataeval/data/_selection.py,sha256=r06xeiyK8nTWPLyItkoPQRWZI1i6LATSue_cuEbCdc4,4463
 dataeval/data/_split.py,sha256=nQABR05vxil2Qx7-uX4Fm0_DWpibskBGDJOYj_b1u3I,16737
 dataeval/data/selections/__init__.py,sha256=2m8ZB53wXzqLcqmc6p5atO6graB6ZyiRSNJFxf11X_g,613
@@ -38,12 +38,12 @@ dataeval/detectors/ood/ae.py,sha256=fTrUfFxv6xUqzKpwMC8rW3JrizA16M_bgzqLuBKMrS0,
 dataeval/detectors/ood/base.py,sha256=9b-Ljznf0lB1SXF4F_Aj3eJ4Y3ijGEDPMjucUsWOGJM,3051
 dataeval/detectors/ood/mixin.py,sha256=0_o-1HPvgf3-Lf1MSOIfjj5UB8LTLEBGYtJJfyCCzwc,5431
 dataeval/metadata/__init__.py,sha256=XDDmJbOZBNM6pL0r6Nbu6oMRoyAh22IDkPYGndNlkZU,316
-dataeval/metadata/_distance.py,sha256=AABrGoQyD13z9Fqlz3NyfX0Iow_vjBwAugIv6OSRTTE,4187
+dataeval/metadata/_distance.py,sha256=MbXM9idsooNWnGLaTKg8j4ZqavUeJUjuW7EPW3-UQyg,4234
 dataeval/metadata/_ood.py,sha256=lNPHouj_9WfM_uTtsaiRaPn46RcVy3YebD1c32vDj-c,8981
 dataeval/metadata/_utils.py,sha256=BcGoYVfA4AkAWpInY5txOc3QBpsGf6cnnUAsHOQTJAE,1210
 dataeval/metrics/__init__.py,sha256=8VC8q3HuJN3o_WN51Ae2_wXznl3RMXIvA5GYVcy7vr8,225
 dataeval/metrics/bias/__init__.py,sha256=329S1_3WnWqeU4-qVcbe0fMy4lDrj9uKslWHIQf93yg,839
-dataeval/metrics/bias/_balance.py,sha256=L5TR8Twwodulk8xkhE-L7PR-isPGw4LusIjL3ZHIH8c,5525
+dataeval/metrics/bias/_balance.py,sha256=fREtoMLUZPOf_ivqNKwij6oPiKMTk02ECO5rWURf3KY,5541
 dataeval/metrics/bias/_completeness.py,sha256=BysXU2Jpw33n5dl3acJFEqF3mFGiJLsfG4n5Q2fkTaY,4608
 dataeval/metrics/bias/_coverage.py,sha256=PeUoOiaghUEdn6Ov8z2-am7-fnBVIPcFbJK7Ty5JObA,3647
 dataeval/metrics/bias/_diversity.py,sha256=25udDKmel9IjeVT5nM4dOa1apda66QdRxBc922yuUvI,5830
@@ -59,7 +59,7 @@ dataeval/metrics/stats/_boxratiostats.py,sha256=ROZrlqgbowkGfCR5PJ5TL7Og40iMOdUq
 dataeval/metrics/stats/_dimensionstats.py,sha256=EVO-BlxrZl8qrP09lwPbyWdrG1ZeDtgj4LiswDwEZ1I,2896
 dataeval/metrics/stats/_hashstats.py,sha256=qa1CYRgOebkxqkALfffaPM-kJ074ZbyfpWbfOfuObSs,4758
 dataeval/metrics/stats/_imagestats.py,sha256=gUPNgN5Zwzdr7WnSwbve1NXNsyxd5dy3cSnlR_7guCg,3007
-dataeval/metrics/stats/_labelstats.py,sha256=bOLH4FEBN4JZ5njdRVjEK7GUb47XBMq9eqYUgXSLmCY,3071
+dataeval/metrics/stats/_labelstats.py,sha256=UG7aKpFctLJvca3rC9sPT_25sCes77KpgZguJYMXfU0,2949
 dataeval/metrics/stats/_pixelstats.py,sha256=5RCQh0OQkHiCkn3DgCPVxKoFfifX_FOtwsnotADSZ0I,3265
 dataeval/metrics/stats/_visualstats.py,sha256=0k6bvAL_d66nQMfG7bydCOFJb7B0dhgG7fqCjVTp1sg,3707
 dataeval/outputs/__init__.py,sha256=geHB5M3QOiFFaQGV4ZwDTTKpqZPvPePbqG7lzaPhaXQ,1741
@@ -67,10 +67,10 @@ dataeval/outputs/_base.py,sha256=-Wa0gFcBVLbfWPMZyCql7x4vGsnkLP4pecsQIeUZ2_Y,590
 dataeval/outputs/_bias.py,sha256=1OZpKncYTryjPLRHb4d6NlhE27uPT57gCob_5jtjKDI,10456
 dataeval/outputs/_drift.py,sha256=rKn5vqMR6XNujgSqfHsH76oFkoGsUusquZL2Qy4Ae6Y,4581
 dataeval/outputs/_estimators.py,sha256=mh-R08CgYtmq9ffANDMYR-V4vrZnSjOjEyOMiMDZ2Ic,3091
-dataeval/outputs/_linters.py,sha256=ZClITD4XY99TunS_9ABTl7eauppoUdpCZU1pCVvD0cI,6700
+dataeval/outputs/_linters.py,sha256=k8lkd8EZ23q0m-HOD-FgqMcLQFy1UH7vws2ucLPyn08,6697
 dataeval/outputs/_metadata.py,sha256=ffZgpX8KWURPHXpOWjbvJ2KRqWQkS2nWuIjKUzoHhMI,1710
 dataeval/outputs/_ood.py,sha256=suLKVXULGtXH0rq9eXHI1d3d2jhGmItJtz4QiQd47A4,1718
-dataeval/outputs/_stats.py,sha256=KIghl-glm9A_h1eVQDKqdTQg8o2zedltWyX4NkCsv2U,15226
+dataeval/outputs/_stats.py,sha256=F-515PGBNB69DXM-YaCkGHAyaXkCD-yYvKfj4-q7R4w,15247
 dataeval/outputs/_utils.py,sha256=NfhYaGT2PZlhIs8ICKUsPWHZXjhWYDkEJqBDdqMeaOM,929
 dataeval/outputs/_workflows.py,sha256=K786mOgegxVi81diUA-qpbwGEkwa8YA7Fk4ttgjJeaY,10831
 dataeval/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -83,7 +83,7 @@ dataeval/utils/_fast_mst.py,sha256=pv42flr1Uf5RBa9qDG0YLDXWH7Mr7a9zpauO1HqZXaY,8
 dataeval/utils/_image.py,sha256=4uxTIOYZZlRJOfNmdA3ek3no3FrLWCK5un48kStMDt8,3578
 dataeval/utils/_method.py,sha256=9B9JQbgqWJBRhQJb7glajUtWaQzUTIUuvrZ9_bisxsM,394
 dataeval/utils/_mst.py,sha256=bLmJmu_1Dtj3hC5gQp3oAiJ_7TKtEjahTqusVRRU4eI,2168
-dataeval/utils/_plot.py,sha256=3yn5UGL2xUeayPNws2bkvxm9ZCURsVUkpvSrwOqUE7g,7145
+dataeval/utils/_plot.py,sha256=1rnMkBRvTFLoTAHqXwF7c7GJ5_5iqlgarZKAzmYciLk,7225
 dataeval/utils/data/__init__.py,sha256=xGzrjrOxOP2DP1tU84AWMKPnSxFvSjM81CTlDg4rNM8,331
 dataeval/utils/data/_dataset.py,sha256=CFK9h-XPN7J-iF2nXol6keMDbGm6VIweFAMAjXRUlhg,9527
 dataeval/utils/data/collate.py,sha256=5egEEKhNNCGeNLChO1p6dZ4Wg6x51VEaMNHz7hEZUxI,3936
@@ -107,7 +107,7 @@ dataeval/utils/torch/models.py,sha256=1idpXyjrYcCBSsbxxRUOto8xr4MJNjDEqQHiIXVU5Z
 dataeval/utils/torch/trainer.py,sha256=Oc2lK13uPGhmLYbmAqlPWyKxgG4YJFlnSXCqFHUZbdA,5528
 dataeval/workflows/__init__.py,sha256=ou8y0KO-d6W5lgmcyLjKlf-J_ckP3vilW7wHkgiDlZ4,255
 dataeval/workflows/sufficiency.py,sha256=j-R8dg4XE6a66p_oTXG2GNzgg3vGk85CTblxhFXaxog,8513
-dataeval-0.86.3.dist-info/LICENSE.txt,sha256=uAooygKWvX6NbU9Ran9oG2msttoG8aeTeHSTe5JeCnY,1061
-dataeval-0.86.3.dist-info/METADATA,sha256=1zOfOabm9w57nxAWZw5InEzmqyWRRko10btPqT0h64o,5353
-dataeval-0.86.3.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-dataeval-0.86.3.dist-info/RECORD,,
+dataeval-0.86.4.dist-info/LICENSE.txt,sha256=uAooygKWvX6NbU9Ran9oG2msttoG8aeTeHSTe5JeCnY,1061
+dataeval-0.86.4.dist-info/METADATA,sha256=qdxTuVh3WxpHvsdRZhAvQIYxiATJLDixoF97xMFYrXM,5353
+dataeval-0.86.4.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+dataeval-0.86.4.dist-info/RECORD,,