The tutor system is being developed for use in conjunction with - not as a substitute for - a few MIT courses. I am a TA for 6.00 (ref: http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-00-introduction-to-computer-science-and-programming-fall-2008/) which is Python geared for non-EECS majors; I'm testing the system with 6.189, a course I've developed for January term (ref: http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-189-a-gentle-introduction-to-programming-using-python-january-iap-2011/). 6.01, "Introduction to EECS", is already utilizing a beta version of the tutor. I think the tutor will work best in conjunction with a class, primarily for students who feel left behind by the instruction. I became a CS major late in the game - my junior year of college - and I struggled with the coursework, as I felt there weren't enough practice problems at the introductory level. The given homework problems often tested advanced understanding of the material, while I struggled to understand the basics.

Like I said, this is being developed for internal MIT use, but I believe that if I spend effort developing this that it should be released openly. Yes, the tutor we have developed is entirely automated. I have spent the summer primarily working on the problems, but as the fall progresses and we work towards developing a partnership with OCW, I hope it will be possible to make the system adaptable for educator use. Perhaps we will be able to enable administrative accounts, or allow educators to mirror the system fully so they can add problems (including non-auto graded ones, which would enable input of open-ended essay or coding questions) as they please, as well as view their students' solutions. I don't know the technical limitations of OCW as of yet, so I can't say what we'll do.

At the very least, I hope to provide practice problems to my students who feel left behind, who often email me desperately asking me for places they can go for simple problems to practice their Python before exams. One thing I hope to do with videos is to provide ones that, after a solution is submitted for a problem, walks through a few different ways of approaching it. I notice that my students often lack a good way of approaching a problem, but when we work through it a second time, they have an easier time grasping what they ought to have looked for and anticipated. I notice students who are willing to spend the time to reconsider their work on an earlier problem often become better, more analytical programmers.

Sarina

On Sun, Aug 28, 2011 at 1:23 AM, Kirby Urner <kurner@oreillyschool.com> wrote:

On Sat, Aug 27, 2011 at 8:31 AM, Sarina Canelake <sarina@mit.edu> wrote:

Hi Carl,

I am a master's student, and for my project I am building an interactive Python tutor system, hopefully with 150-250 problems, as well as embedded video. Some of my inspirations have been the Khan Academy and codingbat.

Heads up re talk at OSCON this year by Scott Gray (full disclosure: I work for the guy) ranting against several categories of interactive courseware, but namely "watch a video then play the match game with canned quiz show type questions". There's no "making with tools", no "apprenticeship". Anyway, just saying there's a spectrum.

This isn't about passing judgment on any particular system, either. Our Python Track has lots of "no brainer" quiz questions for student reinforcement. It's not like this doesn't feel like a school (though more asynchronous and with less emphasis on externally imposed deadlines). The projects tend to be Python modules, or pairs, with one unit testing the other. Teachers provide feedback.

Any time you see complete automation on the back end, I say you have a competitive edge if you add a human touch. Our testing instruments are all hand graded, and projects come with real advisers, albeit not in real time (it's an asynchronous UI).

Kirby

_______________________________________________
Edu-sig mailing list
Edu-sig@python.org
http://mail.python.org/mailman/listinfo/edu-sig

--
Sarina Canelake
MIT EECS