epic-open-source · diehlbw · Dec 2, 2024 · Dec 4, 2024 · Dec 4, 2024 · Dec 4, 2024
@@ -1,6 +1,6 @@
 name: CI
 
-on: 
+on:
   push: {}
   pull_request: {}
   workflow_dispatch: {}
@@ -21,7 +21,7 @@ jobs:
           python-version: "3.10"
       - name: Install pre-commit
         run: |
-          pip install --upgrade pip 
+          pip install --upgrade pip
           pip install pre-commit
       - name: Run pre-commit steps
         run: |
@@ -39,7 +39,7 @@ jobs:
           python-version: "3.10"
       - name: Install dependencies
         run: |
-          pip install --upgrade pip 
+          pip install --upgrade pip
           pip install -e .[dev]
       - name: Run tests
         run: |
@@ -62,8 +62,8 @@ jobs:
   test-packaging:
     name: Test packaging
     runs-on: ubuntu-latest
-  
-    steps: 
+
+    steps:
       - name: Checkout
         uses: actions/checkout@v4
       - name: Setup Python
@@ -72,7 +72,7 @@ jobs:
           python-version: "3.10"
       - name: Install dependencies
         run: |
-          pip install --upgrade pip 
+          pip install --upgrade pip
           pip install build twine
       - name: Build package
         run: |
@@ -84,14 +84,14 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.10", "3.11"]
+        python-version: ["3.10", "3.11", "3.12", "3.13"]
         os:
           - ubuntu-latest
           - windows-latest
           - macos-latest
         include:
           - experimental: false
-      
+
     name: Test python-${{ matrix.python-version }} on ${{ matrix.os }}
     runs-on: ${{ matrix.os }}
     continue-on-error: ${{ matrix.experimental }}
@@ -104,7 +104,7 @@ jobs:
           python-version: ${{ matrix.python-version }}
       - name: Install dependencies
         run: |
-          pip install --upgrade pip 
+          pip install --upgrade pip
           pip install -e .[dev]
       - name: Run tests
         run: |

diff --git a/changelog/100.bugfix.rst b/changelog/100.bugfix.rst
diff --git a/changelog/101.bugfix.rst b/changelog/101.bugfix.rst
diff --git a/changelog/102.feature.rst b/changelog/102.feature.rst
diff --git a/changelog/108.feature.rst b/changelog/108.feature.rst
diff --git a/changelog/109.bugfix.rst b/changelog/109.bugfix.rst
diff --git a/changelog/113.feature.rst b/changelog/113.feature.rst
diff --git a/changelog/114.bugfix.rst b/changelog/114.bugfix.rst
diff --git a/changelog/77.feature.rst b/changelog/77.feature.rst
diff --git a/changelog/86.feature.rst b/changelog/86.feature.rst
diff --git a/docs/release_notes/index.rst b/docs/release_notes/index.rst
@@ -9,6 +9,34 @@ Breaking changes may occur between minor versions prior to the v1 release; after
 
 .. towncrier release notes start
 
+0.3.0
+------
+
+Features
+~~~~~~~~
+
+- Addresses #77 by removing Aequitas and replacing with a great_tables based fairness audit. (`#77 <https://github.com/epic-open-source/seismometer/issues/77>`__)
+- Added ExploreBinaryModelMetrics to see plots of individual metrics, including number needed to treat. (`#86 <https://github.com/epic-open-source/seismometer/issues/86>`__)
+- Reorganize methods making initial import and public api more standard (`#102 <https://github.com/epic-open-source/seismometer/issues/102>`__)
+- Includes confusion matrix rates into Binary Fairness metrics (`#108 <https://github.com/epic-open-source/seismometer/issues/108>`__)
+- Renames `Flagged` to `Flag Rate` for clarity (`#108 <https://github.com/epic-open-source/seismometer/issues/108>`__)
+- Add function to load example datasets (`#113 <https://github.com/epic-open-source/seismometer/issues/113>`__)
+
+
+Bugfixes
+~~~~~~~~
+
+- Remove remaining references to -1 'invalidation'; validate directly on time comparison when needed (`#100 <https://github.com/epic-open-source/seismometer/issues/100>`__)
+- Fixes scaling issue for binary classfier scores that use the range 0-100 rather than 0-1. (`#101 <https://github.com/epic-open-source/seismometer/issues/101>`__)
+- Fixes a few minor ux issues. (`#109 <https://github.com/epic-open-source/seismometer/issues/109>`__)
+
+  - Disable dropdowns with only one valid option.
+  - Fix the Sensitivity/Specificity/PPV plot to move the label to the lower right.
+  - Fix the Legend in the new Fairness Audit table to improve readability.
+  - Add right border to the count column.
+- Remove NotebookHost class that was no longer in use. (`#114 <https://github.com/epic-open-source/seismometer/issues/114>`__)
+
+
 0.2.2
 -----
 

diff --git a/docs/roadmap/index.rst b/docs/roadmap/index.rst
@@ -7,11 +7,11 @@ Roadmap
 Contributions
 =============
 
-As an open-source project, we welcome community contributions to ``seismometer``. 
+As an open-source project, we welcome community contributions to ``seismometer``.
 Ultimately, we want this project to be a community-led effort to codify guidelines
-for ensuring the equitable and informed use of machine learning and AI tools in the 
-healthcare space. Contributions to this project can be as simple as fixing typos or 
-small bugs, or more complex contributions that, with the support and scrutiny of our 
+for ensuring the equitable and informed use of machine learning and AI tools in the
+healthcare space. Contributions to this project can be as simple as fixing typos or
+small bugs, or more complex contributions that, with the support and scrutiny of our
 development team, guide the overall direction of the project.
 
 .. seealso::
@@ -23,20 +23,20 @@ Use Cases
 Templates
 ---------
 
-As of ``v0.1.0``, ``seismometer`` supports evaluating model performance using standardized evaluation 
-criteria binary classifier models. We plan to add support in the near future for other types of 
-machine learning models, such as multiple classifier models. Similarly, we plan to add 
+As of ``v0.1.0``, ``seismometer`` supports evaluating model performance using standardized evaluation
+criteria binary classifier models. We plan to add support in the near future for other types of
+machine learning models, such as multiple classifier models. Similarly, we plan to add
 support for validating generative AI models. These enhancements will include changes to
 the underlying ``seismometer`` tooling, as well as adding new templates for validating
-generative models. 
+generative models.
 
 Workflows and Pre-Live Evaluation
 ---------------------------------
 
 As of ``v0.1.0``, ``seismometer`` has limited support for evaluating model performance pre-live.
-We are planning to add support for workflow simulation (e.g., estimating the number of 
-alerts that would be shown to end-users for a clinical model that predicts an adverse 
-event, or the amount of time saved per clinician for a generative model that drafts 
+We are planning to add support for workflow simulation (e.g., estimating the number of
+alerts that would be shown to end-users for a clinical model that predicts an adverse
+event, or the amount of time saved per clinician for a generative model that drafts
 messages to patients) based on particular thresholds. We will also add tools to identify
 thresholds for models based on pre-live data and operational goals. These tools are intended
 to help identify when a machine learning or artificial intelligence solutions will improve
@@ -45,8 +45,8 @@ current workflows and also improve efficiency when integrating models into a wor
 Comparing to Baselines
 ----------------------
 
-We plan to add support for comparing model performance to baseline statistics (e.g., statistics 
-from a model train or from model performance at a separate site). These are intended to verify 
+We plan to add support for comparing model performance to baseline statistics (e.g., statistics
+from a model train or from model performance at a separate site). These are intended to verify
 that the model feature or target drift are not adversely affecting the model's performance after
 it goes live.
 
@@ -56,14 +56,14 @@ Functional changes
 Visualizations
 --------------
 
-As ``seismometer`` grows, we will add support for new types of visualizations. Our initial focus 
-is to improve visualizations for interventions and outcomes stratified by sensitive groups, but 
+As ``seismometer`` grows, we will add support for new types of visualizations. Our initial focus
+is to improve visualizations for interventions and outcomes stratified by sensitive groups, but
 we plan to extend our model performance visualizations as well.
 
 Data Layer
 ----------
 
-As of ``v0.1.0``, ``seismometer`` supports reading data from `parquet` files, which contain data 
+As of ``v0.1.0``, ``seismometer`` supports reading data from `parquet` files, which contain data
 type information and performance improvements that standard CSV data does not have. We plan to add
 support for more file formats (alongside metadata files that will describe the data types) as well
 as support for reading data directly from a database (e.g., through an ODBC connection).
@@ -77,4 +77,15 @@ which the goal will be to minimize those breaking changes and only release break
 a major version bump.
 
 .. seealso::
-    :ref:`release` for our Release Notes and any breaking changes.
+    :ref:`release` for our Release Notes and any breaking changes.
+
+
+Technical Plans
+===============
+
+Some changes are known to have potential impact on using the package, such as changing supported python
+versions.  As these arise, issues will be created using the `label 'compatibility'`_ for ease of discovery. This
+is distinct from large functional enhancements using a `request for change process`_.
+
+.. _label 'compatibility': https://github.com/epic-open-source/seismometer/labels/compatibility
+.. _request for change process: https://github.com/epic-open-source/seismometer-rfcs
@@ -1,12 +1,12 @@
 [metadata]
 name = seismometer
-version = 0.2.2
+version = 0.3.0
 description = seismometer: Data Science visualization and investigation tools for AI Trust & Assurance
 author = Epic
 author_email = OpenSourceContributions-Python@epic.com
 license_files = LICENSE.txt
 license = BSD 3-Clause
-classifiers = 
+classifiers =
     Programming Language :: Python :: 3
     License :: OSI Approved :: BSD License
     Operating System :: OS Independent
@@ -25,7 +25,7 @@ packages =
     find_namespace:
 include_package_data = True
 install_requires =
-    numpy>=1.26,<2 # >1.26 for python 3.12+
+    numpy>=1.26,<3
     ipython>=8.14
     ipywidgets>=8.0
     jupyterlab>=4.2.5,<5
@@ -37,7 +37,6 @@ install_requires =
     pydantic>=2.6.3,<3
     ydata-profiling>=4.8.3,<5
     great-tables>=0.11.0,<1
-    numba<0.59
 
 [options.packages.find]
 where = src

diff --git a/src/seismometer/data/filter.py b/src/seismometer/data/filter.py
@@ -272,14 +272,14 @@ def notin(cls, column, values) -> "FilterRule":
     @classmethod
     def isna(cls, column) -> "FilterRule":
         """
-        FilterRule where the column contains a na value (np.NaN or None).
+        FilterRule where the column contains a na value (np.nan or None).
         """
         return cls(column, "isna")
 
     @classmethod
     def notna(cls, column) -> "FilterRule":
         """
-        FilterRule where the column does not contain a na value (np.NaN or None).
+        FilterRule where the column does not contain a na value (np.nan or None).
         """
         return cls(column, "notna")
 

diff --git a/src/seismometer/data/performance.py b/src/seismometer/data/performance.py
@@ -77,7 +77,7 @@ def __call__(self, dataframe: pd.DataFrame, metric_names: list[str] = None, **kw
             raise ValueError(f"Invalid metric names: {set(metric_names) - set(self.metric_names)}")
         if len(dataframe) == 0:
             # Universal defaults, if no data frame, return NaN
-            return {name: np.NaN for name in metric_names}
+            return {name: np.nan for name in metric_names}
         full_metrics = self.delegate_call(dataframe, metric_names, **kwargs)
         filtered_metrics = {k: v for k, v in full_metrics.items() if k in metric_names}
         return filtered_metrics

diff --git a/tests/data/test_performance.py b/tests/data/test_performance.py
@@ -211,7 +211,7 @@ def test_generate_metrics_init_correctly(self):
     def test_generate_metrics_empty_dataframe(self):
         metric = undertest.MetricGenerator(["test_metric"], lambda data, names: {"test_metric": 1})
         assert metric.metric_names == ["test_metric"]
-        assert metric(pd.DataFrame()) == {"test_metric": np.NaN}
+        assert metric(pd.DataFrame()) == {"test_metric": np.nan}
 
     def test_generate_named_metrics(self):
         metric = undertest.MetricGenerator(["metric1", "metric2"], lambda data, names: {name: 1 for name in names})