feat: new algorithms + various improvements

.github/actions/validation/action.yml

@@ -3,7 +3,6 @@ description: format, lint, test code
 runs:
   using: "composite"
   steps:
-
     - name: 📦 Install uv
       uses: astral-sh/setup-uv@v6
 
@@ -17,7 +16,7 @@ runs:
 
     - name: 🦾 💅 🧪 Install and validate extras
       run: |
-        extras=("sklearn" "huggingface")
+        extras=("sklearn" "huggingface" "polars" "pandas")
 
         for extra in "${extras[@]}"; do
           echo "🦾 Installing extra: $extra"

Justfile

@@ -7,33 +7,63 @@ set quiet
 default:
   just --list --unsorted
 
+# --------------------------------------------------
+# Developer Setup
+
 # Synchronize the environment by installing all the dependencies
 dev-sync:
     uv sync --cache-dir .uv_cache --all-extras
 
+# Synchronize the environment by installing the specified extra dependency
+# Currently used within the CI to install extra dependencies and test them.
 dev-sync-extra extra:
 	uv sync --cache-dir .uv_cache --extra {{extra}}
 
 # Synchronize the environment by installing all the dependencies except the dev ones
 prod-sync:
-	uv sync --cache-dir .uv_cache --all-extras --no-dev
+	uv sync --cache-dir .uv_cache --all-extras --no-default-groups
+
+# Synchronize the environment by installing the extra dependency
+# specified. Doesn't install the dev dependencies.
+prod-sync-extra extra:
+	uv sync --cache-dir .uv_cache --extra {{extra}} --no-default-groups
 
 # Install the pre-commit hooks
 install-hooks:
 	uv run pre-commit install
 
+# --------------------------------------------------
+# Validation
+
 # Run ruff formatting
 format:
 	uv run ruff format
 
 # Run ruff linting and mypy type checking
 lint:
 	uv run ruff check --fix
-	uv run mypy --ignore-missing-imports --install-types --non-interactive --package ml3_drift
+	uv run mypy --ignore-missing-imports --install-types --non-interactive --package ml3_drift --python-version 3.10
+
+
+# Default value for testWorkers is auto (meaning all workers available)
+# If you want to pass a custom value (such as 4): `just testWorkers=4 test`
+# We also run ruff on tests files (it's so fast that it's worth it)
+
+# Little caveat: when running tests with only an extra installed, you'd like
+# to avoid having docs dependencies installed (since, for instance, a mkdocs plugin
+# requires Pandas, which is one of our extra dependencies). This happens by default
+# since docs dependencies are not installed as default dependencies by uv (see pyproject.toml).
+# They are only installed when building / serving the documentation. However, if you first
+# build the documentation, then run the tests, you will have the docs dependencies installed.
+# Should not be a practical problem (especially since in CI environments we don't install docs dependencies),
+# but it's worth noting.
 
 # Run the tests with pytest
+testWorkers := "auto"
 test:
-	uv run pytest --verbose --color=yes -n auto --exitfirst tests
+    uv run ruff format tests
+    uv run ruff check tests --fix
+    uv run pytest --verbose --color=yes -n {{testWorkers}} --exitfirst tests
 
 # Run linters, formatters and tests
 validate: format lint test
@@ -43,11 +73,12 @@ validate: format lint test
 
 # Generate the documentation
 build-docs:
-	uv run mkdocs build
+    # Make sure mkdocs is installed
+    uv run --group docs mkdocs build
 
 # Serve the documentation locally
 serve-docs:
-	uv run mkdocs serve
+    uv run --group docs mkdocs serve
 
 # --------------------------------------------------
 # Publishing

examples/huggingface/text_embedding_monitoring.py

@@ -3,7 +3,10 @@
 from ml3_drift.huggingface.drift_detection_pipeline import (
     HuggingFaceDriftDetectionPipeline,
 )
-from ml3_drift.huggingface.univariate.ks import KSDriftDetector
+from ml3_drift.monitoring.algorithms.batch.bonferroni import (
+    BonferroniCorrectionAlgorithm,
+)
+from ml3_drift.monitoring.algorithms.batch.ks import KSAlgorithm
 from ml3_drift.callbacks.base import logger_callback
 
 
@@ -37,7 +40,8 @@
     # to monitor the drift in the embeddings.
 
     hf_pipe = HuggingFaceDriftDetectionPipeline(
-        drift_detector=KSDriftDetector(
+        drift_detector=BonferroniCorrectionAlgorithm(
+            algorithm=KSAlgorithm(p_value=0.05),
             callbacks=[
                 partial(
                     logger_callback,

pyproject.toml

@@ -7,19 +7,28 @@ dynamic = ["version"]
 license = { text = "Apache-2.0" }
 readme = "README.md"
 
-dependencies = []
+dependencies = [
+    "scipy>=1.15.3",
+]
 
 # -------------------------------------------------
 # Extra dependencies. This package is designed to be
-# used within one extra at a time, hence we check each
-# extra separately. Remember to update the list of extras
-# in the validation action to ensure tests are run
+# used with different libraries, which means that our code
+# should work only when not all extras are installed.
+# Remember to update the list of extras
+# in the validation CICD to ensure tests are run
 # for your new extra
 [project.optional-dependencies]
 
 sklearn = ["scikit-learn>=1.6.1"]
 
-huggingface = ["scipy>=1.15.2", "transformers[torch]>=4.52.3"]
+huggingface = ["transformers[torch]>=4.52.3"]
+
+polars = ["polars>=1.31.0"]
+
+pandas = ["pandas>=2.2.3"]
+
+river = ["river>=0.22.0"]
 
 
 # -------------------------------------------------
@@ -28,31 +37,45 @@ huggingface = ["scipy>=1.15.2", "transformers[torch]>=4.52.3"]
 dev = [
     "ipykernel>=6.29.5",
     "mypy>=1.15.0",
-    "pillow>=11.2.1",      # for image support in tests
     "pre-commit>=4.1.0",
     "pytest>=8.3.4",
     "pytest-xdist>=3.6.1",
     "ruff>=0.9.5",
-    # for docs
+    # for image support in tests
+    "pillow>=11.2.1",
+]
+
+docs = [
+    "mkdocs-minify-plugin>=0.7.1",
+    "mkdocs-glightbox>=0.3.4",
+    "mkdocs-table-reader-plugin>=2.0.1",
+    "mkdocs-macros-plugin",
     "mkdocs>=1.5.0",
     "mkdocs-material>=9.5.0",
     "mkdocs-material-extensions>=1.1",
     "pygments>=2.14",
     "pymdown-extensions>=9.9.1",
     "jinja2>=3.0",
     "markdown>=3.2",
-    "mkdocs-minify-plugin>=0.7.1",
-    "mkdocs-glightbox>=0.3.4",
-    "mkdocs-table-reader-plugin>=2.0.1",
-    "mkdocs-macros-plugin",
     "openpyxl",
 ]
 
 # -------------------------------------------------
 
+# Default groups for uv
+[tool.uv]
+default-groups = ["dev"]
+
+# -------------------------------------------------
+
 [build-system]
 requires = ["hatchling"]
 build-backend = "hatchling.build"
 
 [tool.hatch.version]
 path = "src/ml3_drift/__init__.py"
+
+
+# Set pytest folder
+[tool.pytest.ini_options]
+testpaths = ["tests"]

ruff.toml

@@ -33,9 +33,6 @@ exclude = [
 line-length = 88
 indent-width = 4
 
-# Assume Python 3.9
-target-version = "py39"
-
 [lint]
 # Enable Pyflakes (`F`) and a subset of the pycodestyle (`E`)  codes by default.
 # Unlike Flake8, Ruff doesn't enable pycodestyle warnings (`W`) or