This repository contains all the resources and artifacts that support the paper entitled "Variability in Data Visualization: a Software Product Line Approach" submmited to the 26th International Systems and Software Product Line Conference (SPLC 2022).
An implementation of the software product line (SPL) for the visualization design process (VDP) that allows generating customized visualizations following the best design practices in data visualization. It contains the following resources classified according to the SPL processes:
- DOMAIN ENGINEERING:
- The feature models in the UVL and FeatureIDE formats that encode the variability of the VDP and knowledge abour best design practices. Concretely, there are four feature models, one for each step in the VDP:
- Step 1: Determine the information you need to display.
- Step 2: Select the best means to display the information.
- Step 3: Design the display to show the information simply, clearly, and accurately.
- Step 4: Select the implementation for the visualization.
- The visualization templates that implement the variability of the visualization artifact. The templates rely on the Jinja2 engine to resolve the variability and generate the final visualization code. Currently, we provide the template for Latex using the pgfplots and tikz packages.
- The mapping models that map the features in the feature model with the visualization templates. Currently, we provide the mapping model for the Latex template using the pgfplots and tikz packages.
- The feature models in the UVL and FeatureIDE formats that encode the variability of the VDP and knowledge abour best design practices. Concretely, there are four feature models, one for each step in the VDP:
- APPLICATION ENGINEERING:
- The configurations of the feature models to generate the visualizations of the four practical scenarios described in the paper. The configurations includes two kinds of files:
- Configurations of the feature models generated with FeatureIDE (.xml). There is at least one configuration for each of the four feature models, and additional configurations for the feature models of Step 3 for each different dataset (i.e., due to the existing multi-feature or clonable feature in the third feature model).
- Configurations of the feature attributes (.csv). There is one file for general attributes in feature model of Step 3, as well as one file each different dataset (i.e., due to the existing multi-feature or clonable feature in the third feature model).
- The Python scripts to orchestrate the all the previous resources and resolve the variability using the Jinj2 template engine.
- The configurations of the feature models to generate the visualizations of the four practical scenarios described in the paper. The configurations includes two kinds of files:
Additionally, we provide the following resources to replicate the visualizations used in the validation:
- The FeatureIDE projects with the feature models and configurations to easy their inspection and editing.
- The quantitative data with the .csv files containing the raw data for the visualization.
- The generated visualizations with our SPL, as well as the adapted visualizations containing the manual modifications to the generated visualizations to fit them in the final paper.
- Python 3.9+ to execute the SPL and generate a visualization.
- FeatureIDE, only in case you want to create new configurations for other visualizations.
To use the SPL follow the next steps:
-
Install Python 3.9+
-
Download this repository and enter into the main directory:
cd spl_visualization_design -
Create a virtual environment:
python -m venv env -
Activate the environment:
In Linux:
source env/bin/activateIn Windows:
.\env\Scripts\Activate -
Install the dependencies:
pip install -r requirements.txt** In case that you are running Ubuntu, please install the package python3.9-dev and update wheel and setuptools with the command
pip install --upgrade pip wheel setuptoolsright after step 4.
To execute the SPL run the following command:
python main.py -f <folder with the configurations>
where the parameter -f indicates the directory that contains all configurations (configurations of feature models and attributes files) for a specific scenario.
See the examples below to replicate the experiments of the paper.
The scripts generate the visualization in the main directory with the source code of the visualization in Latex using the pgfplots and tikz packages. The scripts also print out in the terminal information about the process including:
- The configurations (.xml) found in the directory and considered in the SPL.
- The attributes files (.csv) found in the directory and considered in the SPL.
- The mapping model used.
- Template configuration with the variability resolved.
To replicate the experiments of the paper and generate the visualizations run the following commands:
- Scenario 1:
python main.py -f validation/configurations/scenario1 - Scenario 2:
python main.py -f validation/configurations/scenario2 - Scenario 3:
python main.py -f validation/configurations/scenario3 - Scenario 4:
python main.py -f validation/configurations/scenario4
In all cases, a visualization.tex file is generated in the main directory with the source code of the visualization in Latex using the pgfplots and tikz packages.
In the Validation folder you can find all the resources, including the data files and the generated visualizations.