understand how popular travel modeling packages work by implementing their algorithms from scratch with python

• guiding principles

  • audience: practicing professionals who use travel modeling packages for their work
  • simplicity and clarity: the top priority when implementing the algorithms is that they are easy to understand. Performance considerations should only be a side note or afterthought. Basic programming techniques should be chosen over sophisicated ones.
  • assignment: we only cover the assignment portion of travel modeling. we do not cover demand, distribution, mode choice, etc
  • visualization: algorithms should be illustrated with interesting visualizations, animations, interactions, etc.
  • Python and IPython: algorithms should be implemented in IPthon notebooks, with narrative.
  • references: references for algorithms should be given, but no reference is better than a reference that is not suberbly helpful and well-written. Tutorial-type references should have preference over original references.
  • popular packages: packages that use an algorithm (or some variation of it) should be discussed in the notebook for that algorithm. Emphasis should be given to algorithms used in popular packages used for professional work, rather than experimental or academic approaches.
  • completeness: the note books should help users build an accurate mental model of the entire traffic assignment landscape.
  • from scratch: algorithms should be coded primarily with Python core, with the standard library as needed. Third-party libraries should rarely, if ever, be used. This will serve to dispell the "magic" surrounding commercial travel modeling packages. The one exception to this might be when programming visualizations, animations, etc.

• topics to cover

  • approaches for modeling networks
  • shortest path algorithms
  • static and dynamic assignment
  • micro, meso, and macro assignment