explainiverse 0.1.1a0__tar.gz → 0.1.1a1__tar.gz

explainiverse-0.1.1a1/PKG-INFO

@@ -0,0 +1,128 @@
+Metadata-Version: 2.1
+Name: explainiverse
+Version: 0.1.1a1
+Summary: Unified, extensible explainability framework supporting LIME, SHAP, and custom adapters
+Home-page: https://github.com/jemsbhai/explainiverse
+License: MIT
+Author: Muntaser Syed
+Author-email: jemsbhai@gmail.com
+Requires-Python: >=3.10,<3.13
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Requires-Dist: lime (>=0.2.0.1,<0.3.0.0)
+Requires-Dist: numpy (==1.24.4)
+Requires-Dist: scikit-learn (>=1.1,<1.4)
+Requires-Dist: shap (>=0.48.0,<0.49.0)
+Requires-Dist: xgboost (>=3.0.2,<4.0.0)
+Project-URL: Repository, https://github.com/jemsbhai/explainiverse
+Description-Content-Type: text/markdown
+
+# Explainiverse
+
+**Explainiverse** is a unified, extensible, and testable Python framework for Explainable AI (XAI).  
+It offers a standardized interface for model-agnostic explainability, evaluation metrics like AOPC and ROAR, and support for multiple XAI methods out of the box.
+
+---
+
+## Features
+
+- Standardized `Explainer` API (`BaseExplainer`)
+- Support for:
+  - Local and global feature attribution
+  - Regression and classification tasks
+- Integrated explainers:
+  - **LIME** (tabular, local surrogate)
+  - **SHAP** (KernelExplainer with multi-class, regression, cohort support)
+- Evaluation metrics:
+  - **AOPC** (Area Over Perturbation Curve)
+  - **ROAR** (Remove And Retrain)
+    - Multiple `top_k` support
+    - Baseline options: `"mean"`, `"median"`, `np.ndarray`, `callable`
+    - Curve generation for ROAR vs feature importance
+- Explainability Suite:
+  - Run and compare multiple explainers
+  - Auto-suggestion based on model/task type
+- Built-in support for models: `LogisticRegression`, `RandomForest`, `SVC`, `KNN`, `XGB`, `NaiveBayes`, and more
+
+---
+
+
+## Installation
+
+From PyPI:
+
+```bash
+pip install explainiverse
+```
+
+For development use:
+
+```bash
+git clone https://github.com/jemsbhai/explainiverse.git
+cd explainiverse
+poetry install
+```
+
+---
+
+## Quick Example
+```python
+
+from explainiverse.adapters.sklearn_adapter import SklearnAdapter
+from explainiverse.explainers.attribution.lime_wrapper import LimeExplainer
+from explainiverse.engine.suite import ExplanationSuite
+
+# Wrap your model
+adapter = SklearnAdapter(your_model, class_names=["yes", "no"])
+
+# Build the suite
+suite = ExplanationSuite(
+    model=adapter,
+    explainer_configs=[
+        ("lime", {...}),
+        ("shap", {...})
+    ],
+    data_meta={"task": "classification"}
+)
+
+results = suite.run(instance)
+suite.compare()
+suite.evaluate_roar(X_train, y_train, X_test, y_test, top_k=3)
+```
+
+
+---
+
+## Running Tests
+
+All tests can be run using:
+
+```bash
+poetry run python tests/test_all.py
+```
+
+For individual component testing:
+
+```bash
+poetry run python tests/test_shap_explainer.py
+poetry run python tests/test_lime_explainer.py
+poetry run python tests/test_evaluation_metrics.py
+```
+
+---
+
+## Documentation
+
+Documentation is currently in development.  
+Until then, test files (especially `test_shap_explainer.py`) demonstrate usage and structure.
+
+---
+
+## License
+
+This project is licensed under the MIT License.
+
+

explainiverse-0.1.1a1/README.md

@@ -0,0 +1,105 @@
+# Explainiverse
+
+**Explainiverse** is a unified, extensible, and testable Python framework for Explainable AI (XAI).  
+It offers a standardized interface for model-agnostic explainability, evaluation metrics like AOPC and ROAR, and support for multiple XAI methods out of the box.
+
+---
+
+## Features
+
+- Standardized `Explainer` API (`BaseExplainer`)
+- Support for:
+  - Local and global feature attribution
+  - Regression and classification tasks
+- Integrated explainers:
+  - **LIME** (tabular, local surrogate)
+  - **SHAP** (KernelExplainer with multi-class, regression, cohort support)
+- Evaluation metrics:
+  - **AOPC** (Area Over Perturbation Curve)
+  - **ROAR** (Remove And Retrain)
+    - Multiple `top_k` support
+    - Baseline options: `"mean"`, `"median"`, `np.ndarray`, `callable`
+    - Curve generation for ROAR vs feature importance
+- Explainability Suite:
+  - Run and compare multiple explainers
+  - Auto-suggestion based on model/task type
+- Built-in support for models: `LogisticRegression`, `RandomForest`, `SVC`, `KNN`, `XGB`, `NaiveBayes`, and more
+
+---
+
+
+## Installation
+
+From PyPI:
+
+```bash
+pip install explainiverse
+```
+
+For development use:
+
+```bash
+git clone https://github.com/jemsbhai/explainiverse.git
+cd explainiverse
+poetry install
+```
+
+---
+
+## Quick Example
+```python
+
+from explainiverse.adapters.sklearn_adapter import SklearnAdapter
+from explainiverse.explainers.attribution.lime_wrapper import LimeExplainer
+from explainiverse.engine.suite import ExplanationSuite
+
+# Wrap your model
+adapter = SklearnAdapter(your_model, class_names=["yes", "no"])
+
+# Build the suite
+suite = ExplanationSuite(
+    model=adapter,
+    explainer_configs=[
+        ("lime", {...}),
+        ("shap", {...})
+    ],
+    data_meta={"task": "classification"}
+)
+
+results = suite.run(instance)
+suite.compare()
+suite.evaluate_roar(X_train, y_train, X_test, y_test, top_k=3)
+```
+
+
+---
+
+## Running Tests
+
+All tests can be run using:
+
+```bash
+poetry run python tests/test_all.py
+```
+
+For individual component testing:
+
+```bash
+poetry run python tests/test_shap_explainer.py
+poetry run python tests/test_lime_explainer.py
+poetry run python tests/test_evaluation_metrics.py
+```
+
+---
+
+## Documentation
+
+Documentation is currently in development.  
+Until then, test files (especially `test_shap_explainer.py`) demonstrate usage and structure.
+
+---
+
+## License
+
+This project is licensed under the MIT License.
+

{explainiverse-0.1.1a0 → explainiverse-0.1.1a1}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "explainiverse"
-version = "0.1.1a0"
+version = "0.1.1a1"
 description = "Unified, extensible explainability framework supporting LIME, SHAP, and custom adapters"
 authors = ["Muntaser Syed <jemsbhai@gmail.com>"]
 license = "MIT"

explainiverse-0.1.1a0/PKG-INFO

@@ -1,84 +0,0 @@
-Metadata-Version: 2.1
-Name: explainiverse
-Version: 0.1.1a0
-Summary: Unified, extensible explainability framework supporting LIME, SHAP, and custom adapters
-Home-page: https://github.com/jemsbhai/explainiverse
-License: MIT
-Author: Muntaser Syed
-Author-email: jemsbhai@gmail.com
-Requires-Python: >=3.10,<3.13
-Classifier: License :: OSI Approved :: MIT License
-Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3.10
-Classifier: Programming Language :: Python :: 3.11
-Classifier: Programming Language :: Python :: 3.12
-Requires-Dist: lime (>=0.2.0.1,<0.3.0.0)
-Requires-Dist: numpy (==1.24.4)
-Requires-Dist: scikit-learn (>=1.1,<1.4)
-Requires-Dist: shap (>=0.48.0,<0.49.0)
-Requires-Dist: xgboost (>=3.0.2,<4.0.0)
-Project-URL: Repository, https://github.com/jemsbhai/explainiverse
-Description-Content-Type: text/markdown
-
-# Explainiverse
-
-Explainiverse is a unified, extensible, and testable Python framework for explainable AI (XAI).  
-It provides a consistent API and support for post-hoc explainers like LIME and SHAP, model adapters, and rigorous evaluation strategies.
-
----
-
-## Features
-
-- Standardized Explainer interface (`BaseExplainer`)
-- Support for classification, regression, and multi-class models
-- Integrated explainers:
-  - LIME (Local surrogate models)
-  - SHAP (KernelExplainer with per-class and global support)
-- Adapter layer for scikit-learn models
-- Explanation object with structured output and future extensibility for `.plot()`
-- Full unit test suite covering classification, regression, global/cohort SHAP, and adapter behavior
-
----
-
-## Installation
-
-This package will soon be available on PyPI.
-
-For development use:
-
-```bash
-git clone https://github.com/YOUR_USERNAME/explainiverse.git
-cd explainiverse
-poetry install
-```
-
----
-
-## Running Tests
-
-All tests can be run using:
-
-```bash
-poetry run python tests/test_all.py
-```
-
-For individual component testing:
-
-```bash
-poetry run python tests/test_shap_explainer.py
-poetry run python tests/test_lime_explainer.py
-```
-
----
-
-## Documentation
-
-Documentation is currently in development.  
-Until then, test files (especially `test_shap_explainer.py`) demonstrate usage and structure.
-
----
-
-## License
-
-This project is licensed under the MIT License.
-

explainiverse-0.1.1a0/README.md

@@ -1,61 +0,0 @@
-# Explainiverse
-
-Explainiverse is a unified, extensible, and testable Python framework for explainable AI (XAI).  
-It provides a consistent API and support for post-hoc explainers like LIME and SHAP, model adapters, and rigorous evaluation strategies.
-
----
-
-## Features
-
-- Standardized Explainer interface (`BaseExplainer`)
-- Support for classification, regression, and multi-class models
-- Integrated explainers:
-  - LIME (Local surrogate models)
-  - SHAP (KernelExplainer with per-class and global support)
-- Adapter layer for scikit-learn models
-- Explanation object with structured output and future extensibility for `.plot()`
-- Full unit test suite covering classification, regression, global/cohort SHAP, and adapter behavior
-
----
-
-## Installation
-
-This package will soon be available on PyPI.
-
-For development use:
-
-```bash
-git clone https://github.com/YOUR_USERNAME/explainiverse.git
-cd explainiverse
-poetry install
-```
-
----
-
-## Running Tests
-
-All tests can be run using:
-
-```bash
-poetry run python tests/test_all.py
-```
-
-For individual component testing:
-
-```bash
-poetry run python tests/test_shap_explainer.py
-poetry run python tests/test_lime_explainer.py
-```
-
----
-
-## Documentation
-
-Documentation is currently in development.  
-Until then, test files (especially `test_shap_explainer.py`) demonstrate usage and structure.
-
----
-
-## License
-
-This project is licensed under the MIT License.