Mental models used in introductory data science lessons.
All material made available under the Creative Commons - Attribution license.
Meghan Sposato, Brendan Cullen
Florian Schmoll
Ildiko Czeller, Graham Parsons
Gabriela Sandoval, Florencia D'Andrea, Yanina Bellini Saibene
-
All diagrams are created with diagrams.net, which can be used online or downloaded from the project's GitHub repository (approximately 100MB).
-
Save all diagrams as "Editable SVG" (a superset of the SVG standard that includes information about anchor points and connections).
-
Use the "Sketch" style, which can be chosen from the "Style" menu when no elements are selected. (See this blog post for details.)
-
Use #FEFEFF (very pale blue) as a background color and #000080 (very dark blue) for lines and text. Use 1pt lines, no fill, and 12pt Helvetica. If there is a central concept, give it a 3pt outline and make the text bold and italic
-
Connect both ends of lines to anchor points on concepts and move text as needed so that it does not lie on the line.
-
Export PNG by selecting everything and then using these settings:
- Zoom: 100%
- Border Width: 4 pixels
- Selection Only
- Crop
-
Do not select "Transparent Background", "Shadow", "Embed Images", or "Include a copy of my diagram".
-
Why concept maps? They help authors figure out what they want to teach, give other instructors a quick overview of a lesson, and allow learners to check that they have constructed the right mental model.
-
What about translations? We hope eventually to be able to store translated text in the
.drawio
files. For now, please copy an existing diagram and save it in the appropriate language directory (e.g.,af/topic.drawio
for concept maps in Afar). -
Why SVG? Because vector diagrams rescale much more smoothly than raster images.
-
Why not use color? Or more shapes? Or...? The simpler these diagrams are, the easier they will be to maintain. Simple diagrams also display better on smaller or older devices.
-
Why not use Graphviz, DiagrammeR, or some other tool that compiles text into diagrams?
- As these graphviz-generated files show, it's hard to get their automatic layout algorithms to do as good a job as a person can do.
- They have a much steeper learning curve, which means fewer people will be able to contribute.
- They don't actually make version control easier (reading and understanding the textual description of a diagram is as hard as viewing two diagrams side by side).