Open-Source Curricula for Self-Study & Collaboration

[muraii]

A few months ago I asked for recommendations for textbooks on generalized linear models. (Nevermind that I seem not to have actually asked anything.) I'm still in the same boat but am considering a different approach.

I mention in that I felt approaching practical (mostly work) and theoretical problems on my own was like "a random walk through the space of modelling techniques and undergirding theory." I've looked into graduate programs and tried to form working relationships with potential mentors, but my schedule and rural home make this somewhat difficult.

I'm thinking of going a different route. Imagine a GitHub repo consisting of book titles together with an interesting selection of problems, decided upon by the user group. If the initial repo focused on statistics, and built on a sequence from introductory probability to descriptive and inferential statistics to regression and generalized linear models and on to stochastic processes and Markov chains, this could serve a segment of the interested population. Several books could be detailed with a recommended order both of books and of chapters. The problem selection could be contributed to and decided upon by users via pull requests and the like. Maybe it includes hints for some of the harder problems, or links to relevant documentation.

This achieves an experiential parity for users. You and I, who happen to be working on the same section of the same book, can work together as we choose. A common set of materials helps us help each other and ourselves. It also stands as a resource for many of the questions you find in this forum regarding self-study plans and book recommendations. Book recommendations help, but having a body of other students available to mentor or share approaches improves upon that.

With this framework defined, the project could be forked for a different goal or curriculum. If someone wanted to include some of the texts but rather than progressing through statistics wanted to build out a plan for studying statistical mechanics or differential equations, that's easy enough. No work is wasted or duplicated.

I'm going to start fleshing some of this out for my own planning but if this strikes anyone's interest we could discuss how to make something like this work. The motivation is selfish: for any topic I can find five eagerly recommended books but developing a way of using the book, and a larger plan for progressing beyond the book, isn't so straightforward. By definition I don't know what I'm doing.

Maybe something like this already exists. I know there are the likes of Coursera, Udacity, etc., and I will be looking at them some more. I haven't found very much beyond the basic undergraduate work, however. This potentially allows for any arbitrary book, of any sophistication, to be made a shared learning experience in an ad hoc fashion.

Cheers.

 

[Greg Bernhardt]

Thanks for the post! Sorry you aren't generating responses at the moment. Do you have any further information, come to any new conclusions or is it possible to reword the post?

 

[Doug Huffman]

Unfortunately the best textbooks are not Open Access. My favorite in the field that you may - seems to me - be looking at cost me US$85 for Edwin Thompson Jaynes' Probability Theory: The Logic of Science (Cambridge 2003). I would love to read more deeply in Bayesianism but it is expensive!

My core reading for some years now is Karl Popper. I had to wait two years for the one volume edition of 'Open Society', and am piece-mealing his 'Postscripts' one volume at a time, with one 200 page volume US$65.

MOOCS Coursera pimp the instructor's syllabus/coursebook at obscene prices.

Learn to use arXiv and SSRN

https://en.wikipedia.org/wiki/Open_Access_Button

Greg, can I expect an explanation of a disappeared post of mine, in the 'fever' thread?

 

[muraii]

Thanks for the post! Sorry you aren't generating responses at the moment. Do you have any further information, come to any new conclusions or is it possible to reword the post?

Thanks, Greg. Nothing new just yet, though I realize the proof is in the pudding. I can reword certainly, but I think two nigh-responseless posts on a similar topic point to the need for a new approach. In the spirit of the doocracy, I'm going to start work on an example--using Gelman's text--and see if it progresses how I think, see if it's actually useful to anyone. If so I'll see if it's a generalizable concept and go from there.

Unfortunately the best textbooks are not Open Access. My favorite in the field that you may - seems to me - be looking at cost me US$85 for Edwin Thompson Jaynes' Probability Theory: The Logic of Science (Cambridge 2003). I would love to read more deeply in Bayesianism but it is expensive!

My core reading for some years now is Karl Popper. I had to wait two years for the one volume edition of 'Open Society', and am piece-mealing his 'Postscripts' one volume at a time, with one 200 page volume US$65.

MOOCS Coursera pimp the instructor's syllabus/coursebook at obscene prices.

I certainly think the economics of the publishing machine make this approach challenging, but the fact is these books are available at various libraries with, in my experience, fairly liberal borrowing allowance. I've had Gelman's book checked out for a few months, have renewed it five times, and will continue to until I feel it's time to return it. Sure it requires more diligence than just buying the book but I think it's worth it. I'm going to see if others do, too.

Learn to use arXiv and SSRN

https://en.wikipedia.org/wiki/Open_Access_Button

[PLAIN]https://en.wikipedia.org/wiki/Open_Access_Button[/PLAIN]

Good point, but the issue here is I expect it's easier to find sufficient interest in working through a book commonly referenced and suggested to learners of many stripes than for pre-prints of myriad papers and technical articles. If something like this pans out, perhaps this can be built out, too.