Polar-Diagrams-Dashboard

Manuscript

This library is created for the following paper:

"A Multi-Technique Approach for Improving Summary Polar Diagrams" by Aleksandar Anžel, Zewen Yang, and Georges Hattab

Please cite the paper as: latex Placeholder

Abstract:

While the polar system may lack the universal familiarity of its Cartesian counterpart, it remains indispensable for certain tasks. Summary polar diagrams, such as Taylor and mutual information diagrams, address tasks like discovering relationships, visualizing data similarity, and quantifying correspondence. Although these diagrams are invaluable tools for uncovering data relationships, their polar nature can hinder intuitiveness and lead to issues like overplotting. We present a hybrid approach that combines overview+detail, aggregation, interactive filtering, Cartesian linking, and small multiples to enhance the clarity, comprehensiveness, and functionality of summary polar diagrams. We performed a user study to assess this approach's effectiveness, noting comparable response times among participants. Additionally, three domain experts with varying visualization experience reviewed an implemented solution applying summary polar diagrams to climate, data science (novel), and machine learning, refining the approach prior to the user study. The findings underscore the versatility of our approach in enhancing comprehension, accessibility, and utility.

Dependencies

The code is written in Python 3.11.8 and tested on Linux with the following libraries installed:

|Library|Version| |---|---| |numpy|2.1.0| |pandas|2.2.2| |scikit-learn|1.5.1| |scipy|1.13.1| |polardiagrams|1.2.0| |plotly|5.22.0| |dash|2.14.2| |dashbootstrap_components|1.6.0| |dash-tools|1.12.0| |gunicorn|21.2.0|

The dependencies can also be found in requirements.txt.

Data

|Location|Description| |---|---| |data/|contains all data sets used in the dashboard. |data/CaseStudyClimate/|contains the data set used in the case study "6.1. Climate Model Comparison" in the original paper [1]. |data/CaseStudyEcoli/|contains the data set used for testing the dashboard. |data/CaseStudyWine/|contains the data set used in the case study "6.2. Machine Learning Model Comparison" in the original paper. |data/CaseStudyGaussian_Processes/|contains the data set used in the case study "6.3. Machine Learning Hyper-parameter Tuning" in the original paper. | | |User_Study/data/|contains all data sets generated by the user study and used for the statistical analysis. |UserStudy/data/ResultsFigures/|contains all figures presented in the original paper.

[1] Notes on how to download climate data * The script used for downloading the data/CaseStudyClimate/ was generated using the tutorial found here https://esgf.github.io/esgf-user-support/faq.html#how-to-preserve-the-directory-structure * The script for CMIP5 model data can be automatically re-generated and downloaded using the following link https://esgf-data.dkrz.de/esg-search/wget?downloadstructure=model&project=CMIP5&experiment=historicalExt&variable=ta&ensemble=r2i1p1&timefrequency=mon. * The script for observed (reference) data can be automatically re-generated and downloaded using the following link https://esgf-data.dkrz.de/esg-search/wget/?distrib=false&dataset_id=obs4MIPs.NASA-JPL.AIRS-1-0.mon.ta.gn.v20110608|aims3.llnl.gov.

Code

|Source Code|Description| |---|---| |src/|contains all source scripts. |src/app.py|contains the main script used to build the dashboard. |src/pages/small_multiple.py|contains the script that builds the page with the small multiple technique presented using the data/CaseStudyGaussian_Processes/ data. |src/pages/overview_detail.py|contains the script that builds the page with the overview+detail technique presented using the data/CaseStudyClimate/, data/CaseStudyEcoli/, and data/CaseStudyWine/ data. | | |User_Study/src/|contains the main ipython notebook used to reproduce the results presented in the original paper.

Running

Locally

We recommend downloading the repository and running the dashboard locally due to more responsive interactions. After installing the dependencies from the requirements.txt, the user should run the following commands (Linux):

bash cd Polar-Diagrams-Dashboard/src python app.py

The user should then open the link shown in the terminal or open the browser and type the following address: http://127.0.0.1:8050.

Online

The dashboard is also available online at: https://polar-diagrams-dashboard.onrender.com/. The online version might not be as responsive which is why we recommend running the dashboard locally using the previously mentioned method.

License

Licensed under the GNU General Public License, Version 3.0 (LICENSE or https://www.gnu.org/licenses/gpl-3.0.en.html)

Contribution

Any contribution intentionally submitted for inclusion in the work by you, shall be licensed under the GNU GPLv3.