Scikit-Learn

I have been having a lot of fun in 2013 using scikit-learn with the ipython notebook interface. It seems that most of the things that drive me bonkers about R are easier in scikit-learn. The ipython notebook makes a beautiful IDE that allows for interactive open science notebook development. It allows for easy notes, videos, links, pictures.
Meanwhile scikit-learn allows for a nice, consistent set of interfaces for most supervised, unsupervised and feature selection tasks. The tutorials are great, but even better are the online videos from pycon2013 and SciPy2013.
http://scikit-learn.org/stable/

Data Visualisation

This past weekend I entered my first hackathon, Data In Sight, and our team won!  Data In Sight was a data visualisation competition.  I entered to improve my web development and visualization skills.  Our team, Disaster Masters, investigated the evolving awareness of disasters from 1900-2010.  Thanks to all my teammates.

http://disaster.mmx-dns.com/

http://datainsightresults.wordpress.com/2011/06/26/team-13/

Learning R More Deeply

I have been using R for years, but recently I have been working on a few Data Science problems that are natural in R.  As a result, I have been getting a lot better and utilizing mostly online resources to improve my R skills.  This has been fun and I am getting a glimpse of what the language can do and what it just does not do (data sets bigger than a few Gigs).  Here are a few links that I have been enjoying.

This is a great practical journey into getting things done in an applied field.  the fact that it is psychology makes almost no difference.   http://jeromyanglim.blogspot.com/2009/06/learning-r-for-researchers-in.html

The awesome Joe at Revolution Analytics has put together a great description of the process of learning R and some great tips.  http://www.inside-r.org/howto/how-learn-r

I would add more but these are such a great start and they have linked to all my other favorites.

Online Courseware Rocks

As a motivated self learner, I adore MIT Open Courseware, Stanford Online and iTunes U.  Through MIT, I have studied Python, Algorithms, Stochastic Processes and Optimization.  I am currently deep in professor Ng’s machine learning class.  As a senior technical professional, this is an invaluable educational service provided by these universities.  I wish my almae mater (is that like attorneys general?) participated in this growth of modern education.  Friends in academia, please consider how much putting your courses online can do for your profile in industry and your consulting opportunities.

GMO, commodities and China

The always thoughtful and usually right (though often too early) Jeremy Grantham has a very interesting piece on China’s dominance in the consumption of commodities and what it means for institutional investors.  His thesis is that a slowdown in China will hit commodity indices in a way that will be unexpectedly painful.  While there  is a hint of truth to this, China is a small consumer of oil.  When investors talk about commodities, they may think of pig bellies, but they are really investing in oil.  Goldman’s index is roughly 60% and my preferred, the DJ commodity index pegged near 25%, oil is still the overwhelming beast.

So while this piece is interesting as always, it does not cause nearly the alarm that many of Mr. Grantham’s other posts do.  Yes, a slowdown in China will coincide with commodity trouble, but that merely reflects that institutional investors have dramatically piled into commodities without considering the implications of macro-consistency.  Yes, futures rolling commodity strategies are be very liquid and diversifying now.  Yes, they have had good returns as billions of dollars pile into the strategies.  However, when the shit next hits the fan, there will be some ugly marking to market even there.  It looks like an uncorrelated diversifier, but when the cost of capital is tied to liquidity and futures and liquid hot money, watch out.  But that is true of everything in an unreformed market.  China is not the magic bullet here, and a slowing China will not hit commodities much worse than it will hit the US equity markets.

My first post

Well, you have to start somewhere.  So let’s start where we always do.

Hello World!