http://billmill.org/My Name RhymesBill Mill blogs irregularly2005-04-05T23:13:00ZBill Millbill.mill@gmail.comhttp://billmill.org/Python vs. Illinoishttp://billmill.org/illinois.html2005-04-05T23:13:00ZNo, I'm not dead
yet, I've just been hibernating. And by "hibernating", I mean "watching
college basketball". As the geekier of you probably don't know, the Illinois
men's basketball team just ended a dream season tonight with a loss to North
Carolina in the NCAA tournament finals. Prior to tonight's loss, they only had
<a href="http://www.msnbc.msn.com/id/7108724/">one loss</a>, on a buzzerbeater,
to Ohio State.<p>
"What in the world", I can hear you asking, "does that have to do with Python?"
Well, on a <a href="http://www.yocohoops.com">college basketball board</a> I
frequent, a user named madurzak
<a href="http://www.yocohoops.com/story/2005/4/2/10954/76021">proved
conclusively</a> that the worst team, <a
href="http://en.wikipedia.org/wiki/Ratings_Percentage_Index">RPI-wise</a>, in
college basketball is better than Illinois:<p>
<blockquote>Maryland Eastern Shore beat<br>
MD Baltimore Country which beat<br>
Maine who beat<br>
Northeastern who had a nice win against<br>
UMass who somehow beat<br>
UConn who beat<br>
Indiana who beat<br>
Ohio St. who shocked<br>
Illinois</blockquote><p>
This got me to wondering what some other teams' Illinois Number (as I christened
it, after the famous <a href="http://en.wikipedia.org/wiki/Erdos_number">Erdos
Number</a>) would be. Armed with <a href="http://kenpom.com">Ken Pomeroy's</a>
(<a href="#illinois_txt_foot1">[1]</a>)
<a href="http://kenpom.com/cbbga.txt">game data</a>, Python, and a spare hour,
I set out to find out. Would every team have an Illinois Number? What would the
highest Illinois Number be?<p>
To figure out, I simply built a tree with Illinois at the top, Ohio State below,
the teams that beat Ohio State below that, and so on until all the games that
fit into the tree had been properly placed. I've posted the basic code to
construct the tree <a
href="http://billmill.org/static/files/gen_illinois_tree.py">here</a>; it
requires the <a href="http://kenpom.com/cbbga.txt">data</a> I mentioned
earlier. After some massaging of the results, I threw up a quick CGI app to
explore them <a
href="http://llimllib.f2o.org/illinois_number/illinois.html">here</a>.<p>
<h2>Conclusion</h2><p>
There's no deep conclusion; I just thought it was a neat little script, so I
wanted to post it. There are a few interesting thoughts I had on the data,
though:
<ul><li>The highest Illinois Number is 7, and it belongs to the
little Division III squad of <a
href="http://www.hsc.edu/athletics/basketball/">Hampden Sydney</a>, an
all-men's school in Virginia. They <a
href="http://llimllib.f2o.org/illinois_number/illinois.py?team=hampden+sydney">beat</a>
the <a
href="http://www.longwoodlancers.com/index.asp?path=mbasket">Longwood
Lancers</a>, who managed only one win all season.
<li>Both Division I one-win teams, <a
href="http://llimllib.f2o.org/illinois_number/illinois.py?team=army">Army</a>
and the aforementioned <a
href="http://llimllib.f2o.org/illinois_number/illinois.py?team=longwood">Longwood</a>,
had Illinois Numbers of 6
<li>As far as I can tell, the only Division I school without an Illinois Number
is lowly Savannah State, who went 0-28 this season.
<li>For this type of task - parsing text, building a relatively small tree, and
playing with it - Python is the best. I love playing with data in the
interpreter, coding very very high-level data structures in minutes, and trying
things out as fast as I can type them.<p></ul>
<a name="illinois_txt_foot1">[1]: Let me mention that, if you're a stats geek,
you should have been reading <a href="http://kenpom.com">Ken's site</a> for a
while. Go there for his interesting custom college basketball ranking system,
stay for the game analysis on his blog.<p>
[2]: The data analyzed here does not include tonight's North Carolina game,
since I coded everything this afternoon. Tonight was the first time I was
rooting for a sports team just to keep my blog consistent, but NC's defense was just too good for the Fightin' Illini.
No, I'm not dead
yet, I've just been hibernating. And by "hibernating", I mean "watching
college basketball". As the geekier of you probably don't know, the Illinois
men's basketball team just ended a dream season tonight with a loss to North
Carolina in the NCAA tournament finals. Prior to tonight's loss, they only had
<a href="http://www.msnbc.msn.com/id/7108724/">one loss</a>, on a buzzerbeater,
to Ohio State.<p>
"What in the world", I can hear you asking, "does that have to do with Python?"
Well, on a <a href="http://www.yocohoops.com">college basketball board</a> I
frequent, a user named madurzak
<a href="http://www.yocohoops.com/story/2005/4/2/10954/76021">proved
conclusively</a> that the worst team, <a
href="http://en.wikipedia.org/wiki/Ratings_Percentage_Index">RPI-wise</a>, in
college basketball is better than Illinois:<p>
<blockquote>Maryland Eastern Shore beat<br>
MD Baltimore Country which beat<br>
Maine who beat<br>
Northeastern who had a nice win against<br>
UMass who somehow beat<br>
UConn who beat<br>
Indiana who beat<br>
Ohio St. who shocked<br>
Illinois</blockquote><p>
This got me to wondering what some other teams' Illinois Number (as I christened
it, after the famous <a href="http://en.wikipedia.org/wiki/Erdos_number">Erdos
Number</a>) would be. Armed with <a href="http://kenpom.com">Ken Pomeroy's</a>
(<a href="#illinois_txt_foot1">[1]</a>)
<a href="http://kenpom.com/cbbga.txt">game data</a>, Python, and a spare hour,
I set out to find out. Would every team have an Illinois Number? What would the
highest Illinois Number be?<p>
To figure out, I simply built a tree with Illinois at the top, Ohio State below,
the teams that beat Ohio State below that, and so on until all the games that
fit into the tree had been properly placed. I've posted the basic code to
construct the tree <a
href="http://billmill.org/static/files/gen_illinois_tree.py">here</a>; it
requires the <a href="http://kenpom.com/cbbga.txt">data</a> I mentioned
earlier. After some massaging of the results, I threw up a quick CGI app to
explore them <a
href="http://llimllib.f2o.org/illinois_number/illinois.html">here</a>.<p>
<h2>Conclusion</h2><p>
There's no deep conclusion; I just thought it was a neat little script, so I
wanted to post it. There are a few interesting thoughts I had on the data,
though:
<ul><li>The highest Illinois Number is 7, and it belongs to the
little Division III squad of <a
href="http://www.hsc.edu/athletics/basketball/">Hampden Sydney</a>, an
all-men's school in Virginia. They <a
href="http://llimllib.f2o.org/illinois_number/illinois.py?team=hampden+sydney">beat</a>
the <a
href="http://www.longwoodlancers.com/index.asp?path=mbasket">Longwood
Lancers</a>, who managed only one win all season.
<li>Both Division I one-win teams, <a
href="http://llimllib.f2o.org/illinois_number/illinois.py?team=army">Army</a>
and the aforementioned <a
href="http://llimllib.f2o.org/illinois_number/illinois.py?team=longwood">Longwood</a>,
had Illinois Numbers of 6
<li>As far as I can tell, the only Division I school without an Illinois Number
is lowly Savannah State, who went 0-28 this season.
<li>For this type of task - parsing text, building a relatively small tree, and
playing with it - Python is the best. I love playing with data in the
interpreter, coding very very high-level data structures in minutes, and trying
things out as fast as I can type them.<p></ul>
<a name="illinois_txt_foot1">[1]: Let me mention that, if you're a stats geek,
you should have been reading <a href="http://kenpom.com">Ken's site</a> for a
while. Go there for his interesting custom college basketball ranking system,
stay for the game analysis on his blog.<p>
[2]: The data analyzed here does not include tonight's North Carolina game,
since I coded everything this afternoon. Tonight was the first time I was
rooting for a sports team just to keep my blog consistent, but NC's defense was just too good for the Fightin' Illini.