MidsizeBlowfish

Monetate Labs | Data Scientist | Conshohocken, PA (outside Philadelphia) | Full Time | ONSITE

Monetate Labs is currently looking to hire a data scientist. We are a small team (currently two) that are responsible for R&D on machine learning products at Monetate. We shepherd machine learning products through ideation, feasibility research, and development.

Data Scientist: http://monetate.applytojob.com/apply/ubNfk0jm3Q/Data-Scienti...

Ah, right you are! Thanks for correcting my mistake :)

There is also the interesting interpretation of logistic regression as a latent choice/utility model, where unobserved variation in utility is modeled using the logistic distribution. In this interpretation, the sigmoid function arises as the CDF of the logistic distribution.

This interpretation is also nice as it shows the natural relationship between logistic and probit regression. Probit regression arises when the unobserved utility is modeled with a Gaussian distribution.

Eliminating side effects is about a lot more than variable scope. One example would be reading from/writing to a file.

This. She worked to accommodate your dream, have the decency to to swallow your pride for a bit.

I never set out specifically to donate, but every year I end up giving to pit bull rescues, local animal shelters, and Oxfam.

I was just about to say that this is what about four out of the six years of my math PhD have felt like. (And it's not quite done yet...May cannot get here soon enough!)

A PhD worth getting should not involve you paying tuition. Much of the time your tuition is waived and you earn a small stipend (just enough to live frugally off of) teaching/working in a lab.

From a university's perspective, grad students are cheap labor that can eat a lot of sections of low level undergrad classes.

Just noticed that on my iPad, will look into it.

I'll put a follow up on some other sequential tests into my list of blog-posts-to-do and let you know when it happens.

Haha, yes, I am the author. I do have a very formal style, an unfortunate product of too many years in grad school.

The SPRT is a nice toy test, but only useful for point hypotheses. There are generalizations to more realistic composite hypotheses, though.

In addition to the preposterousness of investing more than a company's market cap, I'd be interested in seeing this strategy backtested. Lots of random crap works over short time scales.

This post is fantastic and the paper is fascinating. I particularly enjoyed his writing:

This is why God, in Her wisdom and mercy, gave us the bootstrap.

The http://arxiv.org/ is a pretty good source. You can get an RSS feed for whichever fields you are interested in.

That's great. I had never seen/considered that.

No problem.

OP here, I have actually not read TAOCP at all. It's just a derivation I sometimes use to pass a few minutes while waiting/bored and I thought I'd blog about it.

In the cases where I've had to use Hadoop, this was exactly what happened. We received a huge volume of data daily, which was aggregated by a nightly Hadoop job into a much more manageable amount of data for us to analyze with python/pandas.

Ooops, that was a typo, thanks for catching it!

This interview was for a data scientist position. I am also a mathematician masquerading as a programmer.

Certainly. This interview was for a data-oriented position.

Poster here, I actually had considered adding both of these methods to the end of the post, but decided to keep it focused on the problem itself.

edit

Maybe I'll write about them in a follow-up post.