Sunday 12 February 2012

Chernoff Faces. Not.


I happened to be looking at Chernoff faces - and they don't work for me. At all.


The don't work for Alex Reisner either - but he has done something about it and has a brilliant piece here (http://alexreisner.com/baseball/stats/chernoff) that explains his solution.


At the risk of spoiling it his solution is a set of Reisner faces - which he shows in a small multiple here to show baseball statistics for 2005 National League.

The raw data is as follows;


PCT H HR BB SB

ARI .475 1419 191 606 67

ATL .556 1453 184 534 92

CHI .488 1506 194 419 65

CIN .451 1453 222 611 72

COL .414 1477 150 509 65

FLO .512 1499 128 512 96

HOU .549 1400 161 481 115

LAD .438 1374 149 541 58

MIL .500 1413 175 531 79

NYM .512 1421 175 486 153

PHI .543 1494 167 639 116

PIT .414 1445 139 471 73

SDP .506 1416 130 600 99

SFG .463 1427 128 431 71

STL .617 1494 170 534 83

WAS .500 1367 117 491 45


He has an algorithm to map these into graphic

win pct = cap rotation

hits = mouth shape

home runs = tongue length

walks = left eye size

stolen bases = right eye size


His set of faces are as follows;