Proofread and Integrate self-instructing course

[bricegnichols]

I realize it's a work in progress, but just wanted to help keep track of it in the issues. Should we retain the section numbers from the original text, for easy cross-reference?

Also, in general, what's best practice on adding formulas and figures here?

[gregmacfarlane]

My class project is wrapping up its conversion of this project this week. At that point, I'll merge it over to a separate TF website, and will be looking for volunteers to go through it to fix/proofread and build links to this project from inside the main TF resource.

[gregmacfarlane]

I would add that we conducted this exercise in mlogit for R. This was pretty simple and straightforward, but there is a problem in that mlogit can't handle multiple-level nesting structures, or constraining two parameters to be equal. Bhat and Koppelman do that a lot. We may need to estimate some of these models in Apollo, but Apollo really doesn't play nicely with bookdown or the literate programming format.

[moeckel]

Biogeme by Michel Bierlaire is a nice alternative. Most importantly, it allows you to exogenously set the nesting coefficients, which MLogit doesn't allow you to do.

[gregmacfarlane]

Can biogeme work literately? Kyle Ward had a good idea to build parallel separate python / R implementations. In mlogit you can constrain a parameter to equal a certain value, if that's what you mean.

[gregmacfarlane]

My students made a valiant effort to convert the SIC into the website. There are 12 chapters in the SIC, and here's what they need:

• 1 Intro: Probably pretty minimal proofreading.
• 2 Elements of Choice: Also minimal proofreading
• 3. Utility Theory: More proofreading
• 4. MNL: proofreading, especially on the math
• 5. Data and basic models: proofreading, possibly some cleanup and double-checking
• 6. Specification refinement: More close proofreading and maybe some edits because mlogit cannot constrain parameters as being equal.
• 7. Other trip purposes: We skipped this chapter entirely
• 8. Nested logit: minimal proofreading
• 9. Selecting a preferred nested structure. This will require more work. mlogit cannot estimate hierarchical nests (as far as I can tell), meaning we probably need to estimate these models in Apollo.
• 10. Multiple Maxima: We skipped this chapter.
• 11. Aggregate assessment: We skipped this chapter.
• 12. Recent advances: we skipped this chapter.

I will be migrating the BYU repository to TF Resource next week. At that point, it would be ready for any members of the committee (especially @singletonpa, @moeckel, and Aruna) to help knock these things out.

[moeckel]

@gregmacfarlane Can you say more specifically how we can help here? Not sure what needs to be done at this point.

[gregmacfarlane]

The new repository is at https://github.com/tfresource/modechoice, and the website is up at https://tfresource.github.io/modechoice/

[vivekyadav26]

@gregmacfarlane Let me know if you need any help with the integration. I am not sure if its already been done.