plausibel: September 2011

US income inequality: Census 2010

The latest census data are out, and they don't look very good. Median household income has fallen for the third year in a row. It stands now at levels of 1996, adjusted for inflation. Have a look at this picture, in particular the development of the income distribution, which tells a story of its own. The US is getting more and more unequal.

Graphs put together by the Wall Street Journal.

click for full size

As Raghuram Rajan writes in his book "Fault Lines", stagnating (or declining) levels of living standards are a problem for a democratic system. The median earner is not happy if his standard of living goes down. To make up for non-rising pay, politicians resort to a different strategy: extend credit.

Sep

Sunday Night Frustration or A Proper Rant

I just finished reading "the big short", a rather vivid account of the events surrounding the 2008 financial crisis, as told by Michael Lewis, who is a former employee of a big - deceased - Wall Street bank called Salomon Brothers. I'll have to say more about the book. It's nothing like an academic paper, but a story about real people, and real events. I am mentioning this because there is people out there who dismiss a story because it's "anectodal evidence", and not "data".

Sep

One big R-esource

I've started to fork out the R-related entries to www.r-bloggers.com. It's a great place to learn R and see what others are up to. The blog has a very nice mixture of topics ranging from extremely simple (stuff that I would post) to extremely advanced, to the very vanguard of R development.

flo.

Sep

Polynomial Interpolation with R

As a first step to produce some useable code for spline interpolation/approximation in R, I set out to first do polynomial interpolation to see how I get along. It's not that there is no spline interpolation software for R, but I find it a bit limited. splinefun, for example, can do only 1-dimensional interpolation. interp{akima} can do bicubic splines - i.e. it can fit a spline surface to a set of points over gridded data as for example (x,y,f(x,y)).

Sep

Fix missing dates with R

I have data on user access to a website. This log file (helpdesk log.csv) just contains the date of access, and how many accesses were counted. It would look like this:

Date hits

13-07-2011 2

14-07-2011 1

16-07-2011 3

17-07-2011 4

...

As you can see, for days with no access (like 15-07-2011 for example), there is no entry.

I wanted to draw a graph showing the number of hits over time. plotting this shows the graph below, but it's conditional on there having been at least one hit.