How well all six models matched the 1993-94 final coaches' poll.
Table explanation: The Spearman Correlation Coefficients (SCC) below relate to how well the top 15, 25 and 35 teams, in the final coaches' poll, match with the predictions for the models listed. The '0,1,2' column contains how many top 35 team rankings were only off by 0, 1, or 2 places, and this is followed by the average deviation between the actual ranking and the prediction. ("X" in the large table at the bottom represents the number of wins in the NIT tournament that a team earned.)
SCC15 | SCC25 | SCC35 | 0,1,2 | Average | |
ZP2 | 0.93729 | 0.98231 | 0.93918 | 9,15,3 | 2.0429 |
PR2 | 0.98214 | 0.99154 | 0.94114 | 13,12,4 | 1.8286 |
50T | 0.97857 | 0.99038 | 0.94324 | 12,13,1 | 1.9143 |
LN2 | 0.96071 | 0.98962 | 0.94758 | 14,10,2 | 1.8857 |
ZPF | 0.89643 | 0.96308 | 0.95081 | 12,6,5 | 2.1857 |
MCB | 0.96429 | 0.97385 | 0.91845 | 6,13,4 | 2.6571 |
OCC | 0.95614 | 0.98038 | 0.96061 | 11,9,7 | 1.8571 |
Commentary: All seven models were extremely accurate, with regards to how the top 25 teams were ranked this year, as illustrated by the high SCC values. The SCC-25 for the PR2 model is the highest value found (from 1993 on) with: 13 eleven exact matches (when compared with the final poll), 12 teams were off by 1 position, four were only off by 2 and one more was 3 positions from its final rank (in the top 35). (The SCC-25 values for the 50T - second best overall so far - LN2 and MCB models were also their highest values ever this year, and OCC's produced its second highest; PR2 predicted 29 teams in the top 35 to be within two or fewer places of their final poll position, ZP2 and OCC predicted 27, 50T and LN2 predicted 26, with ZPF and MCB 23.) The PR2 SCC-15 value tied its highest achievement, with ZP2 and MCB experiencing their second highest SCC-15 values; 50T tied for its 3rd highest, while the OCC model had its 4th highest SCC-15 value as well as its SCC-35 value - and its average difference - both were that model's third best so far as well. The ZPF model did match OCC's correct predictions, for the teams who were ranked #7-#10 in the final coaches' poll, but the first four models listed in the table above had #17 and #18 in reverse order, though those four did predict who was #19 correctly, whereas OCC also had all three of them 'pegged' exactly.
The prediction accuracy regarding #26 StLouis was not so strong for ZP2, PR2, 50T and ZPF, and being off by roughly 10 places significantly impacted those four models' SCC-35 values; similar shortcomings were also realized by these four models - and not OCC - for #27 UAB and #30 Cincinnati. The ZPF model's SCC-15 value was lowered by it ranking #11 Boston College as #17 as well as predicting #15 UMASS to be the #11 team, along with other, small variations of which teams were ranked where. The rank for NIT champion Villanova (#31) was only off by one, in comparison to what the ZP2 and PR2 models predicted, with the 50T, OCC and ZPF models ranking them higher, and LN2 and MCB ranking them lower, than that. All seven models predicted fairly accurately where NIT runnerup Vanderbilt was ranked (in the final coaches' poll).
Team | Poll | Wins | ZP2 | PR2 | 50T | LN2 | ZPF | MCB | OCC |
Arkansas | 1 | 6 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
Duke | 2 | 5 | 2 | 2 | 2 | 2 | 2 | 2 | 2 |
Arizona | 3 | 4 | 3 | 3 | 3 | 3 | 3 | 3 | 4 |
Florida | 4 | 4 | 5 | 4 | 4 | 4 | 6 | 5 | 6 |
Purdue | 5 | 3 | 4 | 5 | 5 | 5 | 4 | 4 | 3 |
Missouri | 6 | 3 | 6 | 6 | 6 | 6 | 5 | 7 | 5 |
Connecticut | 7 | 2 | 8 | 8 | 8 | 8 | 7 | 6 | 7 |
Michigan | 8 | 3 | 7 | 7 | 7 | 7 | 8 | 8 | 8 |
NorthCarolina | 9 | 1 | 10 | 10 | 10 | 10 | 9 | 11 | 9 |
Louisville | 10 | 2 | 9 | 9 | 9 | 9 | 10 | 9 | 10 |
BostonCollege | 11 | 3 | 15 | 11 | 12 | 13 | 17 | 12 | 13 |
Kansas | 12 | 2 | 11 | 12 | 11 | 11 | 12 | 10 | 11 |
Kentucky | 13 | 1 | 14 | 15 | 15 | 15 | 13 | 14 | 15 |
Syracuse | 14 | 2 | 13 | 13 | 13 | 14 | 14 | 13 | 12 |
Massaschusetts | 15 | 1 | 12 | 14 | 14 | 12 | 11 | 17 | 14 |
Indiana | 16 | 2 | 16 | 16 | 16 | 16 | 16 | 15 | 16 |
Marquette | 17 | 2 | 18 | 18 | 18 | 18 | 18 | 16 | 17 |
Temple | 18 | 1 | 17 | 17 | 17 | 17 | 15 | 18 | 18 |
Tulsa | 19 | 2 | 19 | 19 | 19 | 19 | 21 | 19 | 19 |
Maryland | 20 | 2 | 20 | 20 | 20 | 20 | 23 | 23 | 22 |
OklahomaSt | 21 | 1 | 21 | 21 | 21 | 21 | 19 | 20 | 20 |
UCLA | 22 | 0 | 24 | 24 | 23 | 23 | 22 | 27 | 26 |
Minnesota | 23 | 1 | 22 | 22 | 22 | 22 | 20 | 22 | 21 |
Texas | 24 | 1 | 23 | 23 | 24 | 24 | 25 | 21 | 23 |
Pennsylvania | 25 | 1 | 27 | 27 | 28 | 26 | 26 | 26 | 24 |
StLouis | 26 | 0 | 36 | 36 | 35 | 35 | 29 | 32 | 32 |
UAB | 27 | 0 | 39 | 39 | 39 | 36 | 37 | 33 | 34 |
California | 28 | 0 | 25 | 26 | 25 | 25 | 24 | 34 | 27 |
WIGreenBay | 29 | 1 | 31 | 32 | 32 | 29 | 31 | 25 | 29 |
Cincinnati | 30 | 0 | 38 | 38 | 38 | 37 | 38 | 36 | 36 |
Villanova | 31 | "5" | 32 | 30 | 26 | 39 | 28 | 39 | 35 |
Virginia | 32 | 1 | 33 | 33 | 33 | 32 | 34 | 46 | 33 |
WakeForest | 33 | 1 | 26 | 25 | 27 | 27 | 27 | 31 | 25 |
Alabama | 34 | 1 | 34 | 34 | 34 | 31 | 33 | 37 | 30 |
Georgetown | 35T | 1 | 30 | 31 | 31 | 30 | 32 | 24 | 31 |
Nebraska | 35T | 0 | 37 | 37 | 37 | 38 | 39 | 40 | 37 |
Vanderbilt | 37 | "4" | 41 | 41 | 41 | 40 | 40 | 41 | 39 |
Bradley | 38T | "2" | 64 | 70 | 69 | 69 | 69 | 64 | 38 |
NewMexicoSt | 38T | 0 | 55 | 54 | 53 | 44 | 44 | 52 | 48 |
Chattanooga | 38T | 0 | 53 | 51 | 51 | 45 | 45 | 59 | 49 |
SIllinois | 41 | 0 | 46 | 46 | 46 | 46 | 46 | 47 | 50 |
MichiganSt | 42T | 1 | 28 | 28 | 29 | 28 | 30 | 29 | 28 |
NewMexico | 42T | 0 | 44 | 44 | 44 | 43 | 43 | 43 | 46 |
Ohio | 42T | 0 | 43 | 43 | 43 | 42 | 42 | 38 | 41 |