The goal of this exercise is to create a visual representation for all the cars in this dataset. Besides visualizing all rows, the visualization should also enable the representation of as much variables (columns) as possible, by using Bertin's visual attributes and Mackinlay's importance ordering as a basis (Belleman, 2022). For each of the variables, it was assigned a data type, order of priority and visual attribute, as displayed in the following table:
| Column | Data Type | Order of Piority | Visual Attribute |
|---|---|---|---|
| mpg | quantity-ratio | 1st | position x |
| horsepower | quantity-ratio | 2nd | position y |
| weight | quantity-ratio | 3rd | position z |
| cylinders | quantity-ratio | 4th | area |
| origin | nominal | 5th | shape |
| year | quantity-interval | 6th | value |
| model | nominal | 7th | - |
In order to tackle this problem, an interactive visualization application was developed using a Python script that takes advantage of the visualization tools from matplotlib library (Sarkar, 2018).
In order to access the visualization, it's necessary to have Python installed with all the modules specified in the requirements file. Afterwards, the visualization can be launched by running plot.py.
On Windows:
pip install -r requirements.txt
python plot.py
The visualization application also allows for user interaction: within the 3D projection, the camera position can be adjusted, by dragging the mouse pointer.
- Belleman, R. (2022). 1. Introduction [PowerPoint Slides]. Scientific Visualization and Virtual Reality, University of Amsterdam.
- Sarkar, D. (2018). The Art of Effective Visualization of Multi-dimensional Data. Towards Data Science. https://towardsdatascience.com/the-art-of-effective-visualization-of-multi-dimensional-data-6c7202990c57