Blogosphere reentry

I've held off on blogging for a while here while I've been at insight, but the program has officially ended. I'm still going into the office as I prepare for interviews, but I'm not actively working on a project right now, and I have a long backlog of things I'd like to talk about here.

Here are some things I'd like to write about:

• Setting up Bayes rule problems: Apparently these textbook problems are asked commonly in interviews. They're all pretty much the same, and I want to show a nice protocol for solving them. Hopefully it will also help you think about non-textbook problems.
• Statistics for probabilists: Coming from a probability background, I find statistics a little maddening. What question are we asking? What assumptions are we making? How good is the answer? What is the equivalent procedure in probability? This will be a multi-part blog, and I'm not quite qualified to write it yet.
• Machine learning comparison: random forest, support vector machines, etc. I want to talk about the kernel trick.
• Regularization: should probably be its own topic. It guards against overfitting and helps us choose a solution in ill-posed problem. Regularization methods have strange names in data science.
• My Insight project (BackTweet Driver): helps people refine their tweets to increase the chance of being retweeted. You can see it at backtweetdriver.com

...and many more ideas, but you get the idea.

See you soon.