Last week, Kendrick Kay was on a visit in Seattle. Kendrick has been collecting a very large dataset of fMRI data (more about that maybe some other day). In this context, he has been thinking about ridge regression. We spent a couple of hours next to a white board working through his thinking. As always with KK, I learned a ton, and it inspired me to go back and read some of the literature on ridge regression. One of the things that strikes me the most is the relationship between PCA and RR and how other decompositions might correspond to other regression methods.

This paper lays things out nicely and is a good start for thinking about something useful to do.

Meanwhile, Nitime 0.8 is out. In September, this project will celebrate a decade of development. This release dropped Python 2 support. But it turns out that you can’t just drop support for something, without consideration for the downstream projects. So Nitime 0.8.1 is also out.