Coursera
Lies, Damn Lies and Statistics
I kind of hated my paper, and I felt the same way about those I reviewed. I felt like we were being all trained to be the same kind of robot, we all said the the same things like 'if we had more data we would have done even more fantastic at not saying much' using references we never read. But we all covered the bases and we'll get a B or whatever.
I got little out of it, all form no substance.
For a moment at the beginning I thought it would actually be interesting to be massively online The seeing each other's code and comments but then it disappeared lost in what seemed like paranoia about cheating or safety or some such.
The best thing about the course is being turned on to R and R.Studio and .Rmd. Particularly .Rmd.
.Rmd seems to hold promise as a way for researchers and students to explore and present information. In the course it seems as though Jeff uses it then turns it into stale powerpoint slides.
I would be great to have an .Rmd wiki where we each have some personal area to write our explorations hit save and have them run and create a sweet looking page. The lectures could live there too in a public space as .Rmd rendered files. If you need to be all plagarism/security-conscious then you could keep student work secret till some time each week. Ah but after that it would be cool to explore each others code, add comments, steal it to run in your own code area.
What is really exciting would be to extend the .Rmd package to be this generalized calculating, code-running, pretty-formatting workspace. It could be the best thing since wikipedia.
EdX and Coursera have an opportunity to be game changers in education. In Data Analysis it seems we have some of the building blocks already. It could be awesome. Instead we got dumb powerpoint and stupid rubrics and typical paranoia about plagiarism. Same old same old.
Tim McKenna
Define the question Define the ideal data set Determine what data you can access Obtain the data Clean the data Exploratory data analysis Statistical prediction/modeling Interpret results Challenge results Synthesize/write up results Create reproducible code
dog
