Monday, October 6, 2014

Atlas of Design II

It is a real honor to be found in the pages of the second volume of the Atlas of Design.  The contributors make up a list of folks whose craftsmanship I really admire, and it is a privilege to be listed alongside those cartographers.
You can read how I made this map, here, and pick up the book to read a probably too-personal recounting of what it meant to me and the difficult context in which it was made.  Oh well, everything is personal -even the maps that nerds make.


Friday, September 26, 2014

Exponential Outbreak

Epidemiologists have a number they use to describe the average number of people each infected person goes on to infect, to describe the rate of growth of an outbreak.  It's called the effective reproduction number, and it looks like this, where "t" is the number of subsequent infections: Rt


An R of 1 means that the outbreak is steady (and still possibly manageable); one person infects one other person, who then infects one other person, and so on.  If, on average, each person proceeds to infect less than one person the outbreak can be expected to diminish.
An R of 2 means that each person infects two other people and the outbreak will double in new infections every incubation period (which, in the case of Ebola, is thought to be around 12 days).
From June to July of this year, the Rt in Liberia was estimated at 1.7 and reports since then have not indicated an improvement.

I have a hard time directly understanding exponential growth when I see a line chart that curves up rapidly.  I mean, I get it, but an upward trending line is a pretty abstract representation of human suffering.  To illustrate, largely for myself, what exponential growth looks like outside of a line chart, I made an image that shows a hypothetical R2 (doubling) with a twelve-day incubation.
I stopped after 180 days, but thereafter is when the numbers become staggering.


Here is a graphic from one month ago comparing the then-smaller outbreak to all other known Ebola outbreaks.

Tuesday, September 23, 2014

The Air Mile Index: NCAA Football Recruiting Footprints

I was wondering about what various NCAA football team recruiting footprints looked like and what that might tell about the programs.  Here is a map of every player's hometown from the top ten ranked programs right now, according to the AP poll (week 5, September 21).
Every player hometown from the top ten NCAA football programs (except those that fall outside of the map, in Hawaii, American Samoa, Australia, and England. Yes, England).

So I came up with a rating system that looked at the hometowns of all the kids on the team and how far away they were from school and called it the...

Air Miles Index.

You take the distance between a player's hometown and his school (then the square root of that, for lots of reasons), and average it out across the whole team.  That resulting unit-less number is a simple quantifiable way of comparing the reach of recruiting between schools, which might then give us fun clues or insights about the programs and their regions.

If a top-ten program had a really tight local footprint of player hometowns (a low Air Miles Index), that would be pretty interesting.  And if a top-ten program drew kids from all over (a high Air Miles Index), that would be interesting, too!  But why?  Here are the ten best programs in the country right now, in order of their Air Miles index.


Oregon. In addition to being a miner of talent in Hawaii, Oregon pulls in heaps of players from the San Francisco and Los Angeles metro areas.  When I told my 13 year old son, Bear, that I was making maps of where players on the ten best teams came from, I wondered if he could guess the one with the biggest overall reach.  He thought about it for three seconds and then said, "Oregon."
Impressed, I asked him why and he said, "Oregon is really famous because they're so good.  And they look really cool."  He may have nailed it.  When hot-shot high school players consider schools, I wouldn't underestimate the draw of looking cool.  Their partnership with Nike, and the parade of uniforms, might be a talent-attracting vortex.


Next in line is Notre Dame.  While their Air Miles index might be smaller than Oregon's, they have the most distributed of players than any other program mapped here.  Distance seems to have little bearing on kids' deciding to strap on the gold helmets.  I wondered if Notre Dame, as a private Catholic school, has a stronger cultural component on recruitment than the largely regional trends of other schools.  Then co-worker and Notre Dame alum, Scott Morrison, pointed out that this distribution illustrates a benefit of ND's strategy of being independent of a conference -they play all over the country and use that travel as a recruiting mechanism.
ND has some interesting clusters.  The team, which is wholly devoid of locals, pretty much comes from LA, Dallas, Miami, Chicago, and Boston.  Talk about spread out.  Also, the Air Miles Index has remained pretty consistent across all of the classes.


Oklahoma plucks a hefty number of players from over the Texas border in Dallas / Ft. Worth, but compared to the other teams in this list, they are flush with home-grown players from just a couple miles away in Oklahoma City.  The breakout across classes shows freshmen and sophomores have a pretty low Air Miles Index compared to the juniors and seniors, indicating the Sooners are getting even more local in their representation.


Florida State has an amazingly in-state roster; not many of these players are crossing a border to come to school.  While their overall Air Miles Index is higher than Alabama (below) they are mostly Floridians.


Alabama has a pretty evenly dispersed team.  While most of the team comes from within the region, there isn't really a distant recruiting hot-spot city, like some other teams.

Baylor, a private Baptist university, has almost the inverse pattern of recruitment than Notre Dame (except for their shared love of Dallas Ft. Worth).  Baylor is almost entirely Texan, with a few big exceptions (Australia and England).  The junior class was an uncharacteristically distant set of recruits.


The University of Mississippi's Air Miles Index, while relatively low, could be much lower were it not for the players piped in from Dallas and Atlanta.


The Spartans of Michigan State (a couple miles down the road from me, as it happens) are pretty much the all-star team of metro-Detroit high schools.  Some players from neighboring states push up the Air Miles Index a bit, but the overall characteristic is that of dense local talent -particularly for the current senior class.


Predominately made up of kids from Birmingham and metro Atlanta, the Tigers don't need to reach very far to put together a top ten team.  Maybe most notable about Auburn's low Air Miles Index is that not a single player comes from Texas!


The Aggies are the lowest Air Miles team in the AP poll's top ten, indicating a density of local talent.  Nestled between high school football jackpots Dallas and Houston, College Station is a goldilocks campus straddling two massive nearby epicenters.  Out-of-towners are rare, indeed.

What about by position?  Since I had the data and had figured the distances (the hard part), it was a quick pivot table to get the average Air Mile Index per position.  Of course I would expect punters to be really distantly recruited (given US football's tendency to court foot-familiar kids raised on soccer and rugby), but I thought marquee positions like quarterbacks, wide receivers, and middle linebackers to be recruited from greater distances, but the top-ten programs tended to stay relatively local for those positions.  They also apparently have a hard time finding local Centers.  I thought that was weird.

PositionAir Mile Index
DT14.1
LS14.7
DE15.3
S15.4
PK16.0
LB16.0
FB16.2
RB16.4
TE16.4
DL17.0
QB17.0
WR17.2
G17.2
DB18.0
CB18.6
OT18.8
OL20.6
C21.4
P28.8


Ok, that's it for the ten top-ranked teams.

The following schools are not currently top-ten AP poll ranked programs (of course I'm confidently holding out for CMU -just a matter of time, really), but I was curious about them so while the process was fresh in my soft soft mind, I thought I'd crank out Air Mile Indexes for Penn State, Michigan, and Central Michigan.

Joshua Stevens was curious about PSU.  The Nittany Lions draw a lot of kids from New Jersey and their Air Mile Index is higher than Michigan State and lower than Ole Miss.  The freshman class' Air Mile Index dwarfs those of the other years, indicating a recent drive to recruit at greater distances.  Also, the number of players across the classes is similar to the trend seen at Notre Dame (above).


The wolverines have had some struggles in recent years, but the Air Mile Index across the four classes is pretty flat -no indication of growing or shrinking reach.


Understandably, Central's Air Miles Index is much lower than the rest of the teams in this list.  As a regional state university, the players come almost entirely from within Michigan except for what looks like a well-traveled recruitment turn in western-suburban Chicago (and a few in Miami).


 

Thursday, September 4, 2014

Without Scotland, What of the Union Jack?

The  referendum on whether Scotland should be an independent country carries with it vast political, social, and economic questions.  Chief among them, of course, is what to do about the Union Jack?