Bayes-Works · van-dai-vuong · Jun 4, 2025 · Jun 4, 2025 · Jun 4, 2025 · Jun 4, 2025
diff --git a/.github/workflows/build_wheels.yml b/.github/workflows/build_wheels.yml
@@ -3,7 +3,7 @@ name: Build and Publish to PyPI
 on:
   push:
     branches:
-      - main
+      - main 
     tags:
       - 'v*'
   pull_request:
@@ -24,25 +24,16 @@ jobs:
         with:
           python-version: "3.10"
 
+      - name: Install system dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y libopenmpi-dev
+
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
           python -m pip install build twine
 
-          # Install pytagi from github (TODO: because cannot install from pip)
-          git clone --branch v0.2.0 https://github.com/lhnguyen102/cuTAGI.git
-          cd cuTAGI
-          git submodule update --init --recursive
-          pip install -r requirements.txt
-          pip install .
-          cd ..
-          # 
-          pip install -e .
-
-      - name: Run unit tests
-        run: pytest canari/test
-        working-directory: ${{ github.workspace }}/..
-
       - name: Clean build environment
         run: rm -rf dist/ build/ *.egg-info
 
@@ -57,3 +48,24 @@ jobs:
         run: |
           twine upload --verbose --repository pypi dist/*
 
+  unit-test:
+    name: Unit Tests on macOS
+    runs-on: macos-latest
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v3
+
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.10"
+
+      - name: Install dependencies
+        run: |
+          pip install pytagi
+          pip install -e .
+
+      - name: Run unit tests
+        run: pytest canari/test
+        working-directory: ${{ github.workspace }}/..
diff --git a/.github/workflows/deploy_docs.yml b/.github/workflows/deploy_docs.yml
@@ -6,17 +6,9 @@ on:
     branches:
       - main
 
-# on:
-#   workflow_dispatch:
-
-# on:
-#   pull_request:
-#     branches:
-#       - main
-
 jobs:
   deploy:
-    runs-on: ubuntu-latest
+    runs-on: macos-latest
 
     steps:
       - name: Checkout code
@@ -31,7 +23,8 @@ jobs:
         run: |
           python -m pip install --upgrade pip
           pip install -r docs/requirements.txt
-          sudo apt-get update && sudo apt-get install -y pandoc
+          pip install -r requirements.txt
+          brew install pandoc
 
       - name: Manually add src/ to PYTHONPATH
         run: echo "PYTHONPATH=$PYTHONPATH:$(pwd)/src" >> $GITHUB_ENV

diff --git a/README.md b/README.md
@@ -1,26 +1,35 @@
 ## What is Canari?
-Canari is an open-source library for online change point detection in univariate time series. It has been developed in the context of long-term online monitoring where we aim for the early detection of subtle change points in the baseline response, while limiting to a minimum the number of false alarms. Because Canari’s probabilistic change-point detection relies on 1-step ahead predictions, it is inherently capable of performing short- and long-term forecasting, as well as online data imputation; One-step ahead predictions can either rely solely on a target time series, or on explanatory time series that can be non-linearly related to the target.
+Canari is an open-source library for online change point detection in univariate time series. It has been developed in the context of long-term online monitoring where we aim for the early detection of subtle change points in the baseline response, while limiting to a minimum the number of false alarms.
+
+Because Canari’s probabilistic change-point detection relies on 1-step ahead predictions, it is inherently capable of performing short- and long-term forecasting, as well as online data imputation; One-step ahead predictions can either rely solely on a target time series, or on explanatory time series that can be non-linearly related to the target.
+
 In addition to change-point detection and forecasting capabilities, it returns an interpretable decomposition of time series into baseline components representing the irreversible responses, recurrent pattern components caused by external effects, and residual components allowing the characterization of both the model prediction and the observation uncertainties.
 
 ## How does it work?
-```{figure} ../docs/_static/Canari_SSM_LSTM.png
----
-scale: 30%
-align: right
----
-```
-The methodological core behind the canary library consists of a seamless integration between [state-space models (SSM)](http://profs.polymtl.ca/jagoulet/Site/PMLCE/CH12.html) and Bayesian neural networks. On the one hand, the Gaussian SSM theory enables modelling baseline responses and residuals, while on the other, [Tractable Approximate Gaussian Inference (TAGI)](https://github.com/lhnguyen102/cuTAGI/tree/main) also enables treating all parameters and hidden states in neural networks as Gaussians. Canari uses LSTM neural networks to model recurrent patterns as well as non-linear dependencies with respect to explanatory variables. Because both the SSM and LSTM rely on the same Gaussian conditional inference mechanism, their hidden states can be inferred analytically in a same unified probabilistic framework.
-The figure on the right presents an example where the raw data in red is decomposed in a baseline that is characterized by a baseline “level” component where its rate of change is described by the “trend”. The recurrent pattern is modelled by a LSTM Bayesian neural network where the training and validation set consists only in 4 years of data. The residual characterizing the model errors is itself modelled by a white “noise” component. The change point detection can be performed either online or offline after the training and validation period; The presence of change points is indicated by the probability of regime switches that rise toward 1 on several occasions.
-```{figure} ../docs/_static/Canari_example.png
-```
+
+<img src="./docs/_static/Canari_SSM_LSTM.png" alt="Canari SSM LSTM" width="23%" align="right">
+
+The methodological core behind the canary library consists of a seamless integration between [state-space models (SSM)](http://profs.polymtl.ca/jagoulet/Site/PMLCE/CH12.html) and Bayesian neural networks. On the one hand, the Gaussian SSM theory enables modelling baseline responses and residuals, while on the other, [Tractable Approximate Gaussian Inference (TAGI)](https://github.com/lhnguyen102/cuTAGI/tree/main) also enables treating all parameters and hidden states in neural networks as Gaussians.
+
+Canari uses LSTM neural networks to model recurrent patterns as well as non-linear dependencies with respect to explanatory variables. Because both the [SSM and LSTM rely on the same Gaussian conditional](https://www.sciencedirect.com/science/article/pii/S0169207024000335) inference mechanism, their hidden states can be inferred analytically in a same unified probabilistic framework.
+
+<img src="./docs/_static/Canari_example.png" alt="Canari example" width="40%" align="right">
+
+The figure above presents an example where the raw data in red is decomposed in a baseline that is characterized by a baseline “level” component where its rate of change is described by the “trend”. The recurrent pattern is modelled by a LSTM Bayesian neural network where the training and validation set consists only in 4 years of data. The residual characterizing the model errors is itself modelled by a white “noise” component. The change point detection can be performed either online or offline after the training and validation period; The presence of change points is indicated by the probability of regime switches that rise toward 1 on several occasions.
 
 ## Getting started
-You can get started with Canari by going through our [installation guide](https://bayes-works.github.io/canari/installation_guide.html) and [tutorials](https://bayes-works.github.io/canari/tutorials.html) covering all the main features of the library.
+You can get started with Canari by going through our [tutorials](https://bayes-works.github.io/canari/tutorials.html) covering all the main features of the library.
 
+## Installation
+Canari available on PyPI, for installation, execute the following command in Terminal:
+```
+pip install pycanari
+```
+For more info. on how to install, visit our [installation guide](https://bayes-works.github.io/canari/installation_guide.html).
 ## Contributors
-The principal developer of canary is Van Dai Vuong with the mentoring of Luong Ha Nguyen and oversight by James-A. Goulet. The major contributors to the library are
-	- Zhanwen Xin (Online AR, Bounded AR & several+++ bug fixes through PRs)
-	- David Wardan (SLSTM component & several+++ bug fixes through PRs)
+The principal developer of canary is Van Dai Vuong with the mentoring of Luong Ha Nguyen and oversight by James-A. Goulet. The major contributors to the library are:
+- Zhanwen Xin (Online AR, Bounded AR & several+++ bug fixes through PRs)
+- David Wardan (SLSTM component & several+++ bug fixes through PRs)
 
 # Acknowledgements
 We acknowledge the financial support from Hydro-Québec and the Natural Sciences and Engineering Research Council of Canada in the development of the Canari library.
@@ -39,8 +48,8 @@ Canari is released under the MIT license.
 * [Anomaly Detection with the Switching Kalman Filter for Structural Health Monitoring](https://profs.polymtl.ca/jagoulet/Site/Papers/2017_Nguyen_and_Goulet_AD-SKF.pdf) (Luong Ha Nguyen and James-A. Goulet, 2018)
 * [Bayesian dynamic linear models for structural health monitoring](https://profs.polymtl.ca/jagoulet/Site/Papers/Goulet_BDLM_SHM_2017_preprint.pdf)(James-A. Goulet, 2017)
 ### Theses
-[Analytically Tractable Bayesian Recurrent Neural Networks with Structural Health Monitoring Applications](https://profs.polymtl.ca/jagoulet/Site/Papers/DV_Thesis_2024.pdf) (Van-Dai Vuong, 2024)
+*[Analytically Tractable Bayesian Recurrent Neural Networks with Structural Health Monitoring Applications](https://profs.polymtl.ca/jagoulet/Site/Papers/DV_Thesis_2024.pdf) (Van-Dai Vuong, 2024)
 * [Analytical Bayesian Parameter Inference for Probabilistic Models with Engineering Applications](https://profs.polymtl.ca/jagoulet/Site/Papers/BhargobDekaThesis.pdf) (Bhargob Deka, 2022)
 * [Real-time Anomaly Detection in the Behaviour of Structures](https://profs.polymtl.ca/jagoulet/Site/Papers/LHNguyen_these_2019.pdf) (Luong Ha Nguyen, 2019)
 ### Book chapter
-* [Probabilistic machine learning for civil engineers (chapter 12 - SSM)](http://profs.polymtl.ca/jagoulet/Site/PMLCE/CH12.html) (James-A. Goulet. 2020)
+* [Probabilistic machine learning for civil engineers (chapter 12 - SSM)](http://profs.polymtl.ca/jagoulet/Site/PMLCE/CH12.html) (James-A. Goulet. 2020)
diff --git a/docs/_static/Canari_SSM_LSTM.png b/docs/_static/Canari_SSM_LSTM.png
diff --git a/docs/_static/Canari_example.png b/docs/_static/Canari_example.png
diff --git a/docs/conf.py b/docs/conf.py
@@ -8,7 +8,7 @@
 project = "canari"
 copyright = "2025, Van-Dai Vuong, Luong-Ha Nguyen, James-A. Goulet"
 author = "Van-Dai Vuong, Luong-Ha Nguyen, James-A. Goulet"
-release = "v.0.0.2"
+release = "v.0.0.3"
 
 extensions = [
     "sphinx.ext.autodoc",
@@ -21,8 +21,6 @@
 ]
 nbsphinx_execute = "never"
 
-autodoc_mock_imports = ["pytagi"]
-
 templates_path = ["_templates"]
 exclude_patterns = []
 

diff --git a/docs/index.rst b/docs/index.rst
@@ -4,19 +4,19 @@ Canari's documentation
 What is Canari?
 ---------------
 
-Canari is an open-source library for online change point detection in univariate time series. 
-It has been developed in the context of long-term online monitoring where we aim for the early 
-detection of subtle change points in the baseline response, while limiting to a minimum the 
+Canari is an open-source library for online change point detection in univariate time series.
+It has been developed in the context of long-term online monitoring where we aim for the early
+detection of subtle change points in the baseline response, while limiting to a minimum the
 number of false alarms.
 
-Because Canari's probabilistic change-point detection relies on 1-step ahead predictions, 
-it is inherently capable of performing short- and long-term forecasting, as well as online 
-data imputation. One-step ahead predictions can either rely solely on a target time series, 
+Because Canari's probabilistic change-point detection relies on 1-step ahead predictions,
+it is inherently capable of performing short- and long-term forecasting, as well as online
+data imputation. One-step ahead predictions can either rely solely on a target time series,
 or on explanatory time series that can be non-linearly related to the target.
 
-In addition to change-point detection and forecasting capabilities, it returns an interpretable 
-decomposition of time series into baseline components representing the irreversible responses, 
-recurrent pattern components caused by external effects, and residual components allowing the 
+In addition to change-point detection and forecasting capabilities, it returns an interpretable
+decomposition of time series into baseline components representing the irreversible responses,
+recurrent pattern components caused by external effects, and residual components allowing the
 characterization of both the model prediction and the observation uncertainties.
 
 How does it work?
@@ -26,27 +26,29 @@ How does it work?
    :scale: 30%
    :align: right
 
-The methodological core behind the Canari library consists of a seamless integration between 
-`state-space models (SSM) <http://profs.polymtl.ca/jagoulet/Site/PMLCE/CH12.html>`_ and Bayesian 
-neural networks. On the one hand, the Gaussian SSM theory enables modelling baseline responses and 
-residuals, while on the other, `Tractable Approximate Gaussian Inference (TAGI) <https://github.com/lhnguyen102/cuTAGI/tree/main>`_ 
+The methodological core behind the Canari library consists of a seamless integration between
+`state-space models (SSM) <http://profs.polymtl.ca/jagoulet/Site/PMLCE/CH12.html>`_ and Bayesian
+neural networks. On the one hand, the Gaussian SSM theory enables modelling baseline responses and
+residuals, while on the other, `Tractable Approximate Gaussian Inference (TAGI) <https://github.com/lhnguyen102/cuTAGI/tree/main>`_
 enables treating all parameters and hidden states in neural networks as Gaussians.
 
-Canari uses LSTM neural networks to model recurrent patterns as well as non-linear dependencies 
-with respect to explanatory variables. Because both the SSM and LSTM rely on the same Gaussian 
-conditional inference mechanism, their hidden states can be inferred analytically in a unified 
+Canari uses LSTM neural networks to model recurrent patterns as well as non-linear dependencies
+with respect to explanatory variables. Because both the `SSM and LSTM rely on the same Gaussian
+conditional <https://www.sciencedirect.com/science/article/pii/S0169207024000335>`_ inference mechanism, their hidden states can be inferred analytically in a unified
 probabilistic framework.
 
 .. figure:: _static/Canari_example.png
+   :scale: 24%
+   :align: right
 
-The figure above presents an example where the raw data in red is decomposed into a baseline that 
-is characterized by a baseline “level” component where its rate of change is described by the “trend.” 
-The recurrent pattern is modelled by an LSTM Bayesian neural network where the training and validation 
-set consists only of 4 years of data. The residual characterizing the model errors is itself modelled 
+The figure on the right presents an example where the raw data in red is decomposed into a baseline that
+is characterized by a baseline “level” component where its rate of change is described by the “trend.”
+The recurrent pattern is modelled by an LSTM Bayesian neural network where the training and validation
+set consists only of 4 years of data. The residual characterizing the model errors is itself modelled
 by a white “noise” component.
 
-Change point detection can be performed either online or offline after the training and validation 
-period. The presence of change points is indicated by the probability of regime switches that rise 
+Change point detection can be performed either online or offline after the training and validation
+period. The presence of change points is indicated by the probability of regime switches that rise
 toward 1 on several occasions.
 
 Getting started
@@ -69,7 +71,7 @@ The major contributors to the library are:
 Acknowledgements
 ----------------
 
-We acknowledge the financial support from Hydro-Québec and the Natural Sciences and 
+We acknowledge the financial support from Hydro-Québec and the Natural Sciences and
 Engineering Research Council of Canada in the development of the Canari library.
 
 License

diff --git a/docs/requirements.txt b/docs/requirements.txt
@@ -1,7 +1,3 @@
 Sphinx>=6.1
 sphinx-book-theme>=0.3.4
 nbsphinx>=0.8.8  
-matplotlib>=3.5
-numpy>=1.22
-pandas
-ray[tune]
diff --git a/pyproject.toml b/pyproject.toml
@@ -4,9 +4,13 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "pycanari"
-version = "0.0.2"
-description = "Open-source library for probabilistic anomaly detection in time series"
-readme = "README.md"
+version = "0.0.3"
+description = """
+Canari is an open-source library for online change point detection in univariate time series. 
+It has been developed in the context of long-term online monitoring where we aim for the early 
+detection of subtle change points in the baseline response, while limiting to a minimum the number
+of false alarms.
+"""
 license = { text = "MIT" }
 requires-python = ">=3.10"
 
@@ -39,4 +43,8 @@ package-dir = { "" = "src" }
 [tool.setuptools.packages.find]
 where = ["src"]
 
+[project.urls]
+Homepage = "https://bayes-works.github.io/canari/index.html"
+Repository = "https://github.com/Bayes-Works/canari"
+
 
diff --git a/test/saved_metric/test_model_forecast_metric.csv b/test/saved_metric/test_model_forecast_metric.csv
@@ -1,2 +1,2 @@
 mse
-0.002291753
+2.1171463e-06
diff --git a/test/_test_model_forecast.py → test/test_model_forecast.py b/test/_test_model_forecast.py → test/test_model_forecast.py