Due to space, we were not able to include complete information on the datasets and models learned within the paper. The table below extends Table 2 from the paper with additional information.
Explanation of column headings:
For cross-validated datasets, numbers listed are the means across all 10 training/test splits. When computing the mean, log likelihoods were properly weighted to give every test example equal weight, even though some cross-validation splits have more test examples than others.
Name | #Vars | Examples | Tuning | Validation | Test | NBE | WinMine | Marginal | Clusters | Splits |
1985 Auto Imports | 26 | 205 | 125.1 | 59.4 | 20.5 | -25.68 ± 0.53 | -24.55 ± 0.42 | -36.16 ± 0.26 | 21 | 169.6 |
Abalone | 9 | 4,177 | 2,773 | 985 | 419 | -7.25 ± 0.13 | -7.27 ± 0.14 | -13.909 ± 0.018 | 51 | 90 |
Adult | 15 | 5,423 | 3,895 | 993 | 535 | -13.00 ± 0.16 | -12.72 ± 0.14 | -15.96 ± 0.14 | 50 | 147 |
Annealing | 39 | 798 | 483.6 | 234.6 | 79.8 | -10.784 ± 0.094 | -10.238 ± 0.090 | -15.08 ± 0.13 | 22.3 | 105.4 |
Anonymous MSWeb | 294 | 37,711 | 29,441 | 3,270 | 5,000 | -9.91 ± 0.10 | -9.69 ± 0.10 | -11.36 ± 0.12 | 50 | 1,416 |
Audiology | 70 | 226 | 138.6 | 64.8 | 22.6 | -16.15 ± 0.38 | -15.65 ± 0.40 | -18.98 ± 0.42 | 10.6 | 85.4 |
Auto MPG | 8 | 398 | 242.1 | 116.1 | 39.8 | -9.17 ± 0.14 | -9.10 ± 0.13 | -12.720 ± 0.042 | 27.8 | 72.6 |
Breast Cancer Wisconsin | 31 | 569 | 343.5 | 168.6 | 56.9 | -36.57 ± 0.29 | -31.33 ± 0.24 | -48.9466 ± 0.0080 | 30.6 | 171.6 |
BUPA | 7 | 345 | 209.6 | 100.9 | 34.5 | -9.862 ± 0.075 | -9.874 ± 0.054 | -10.157 ± 0.031 | 25 | 4.3 |
Car | 7 | 1,728 | 1,037.2 | 518 | 172.8 | -7.8242 ± 0.0086 | -7.705 ± 0.010 | -8.301 ± 0.020 | 50.3 | 35.4 |
Census | 14 | 45,222 | 36,691 | 4,075 | 4,456 | -11.028 ± 0.056 | -10.788 ± 0.050 | -15.161 ± 0.061 | 306 | 316 |
Chess Endgames | 37 | 3,196 | 1,932 | 937 | 327 | -10.79 ± 0.11 | -9.724 ± 0.093 | -15.11 ± 0.18 | 56 | 723 |
Connect-4 | 43 | 67,557 | 54,738 | 6,111 | 6,708 | -15.079 ± 0.015 | -13.902 ± 0.017 | -20.629 ± 0.039 | 328 | 7,954 |
Contraceptive Method Choice | 10 | 1,473 | 885.3 | 440.4 | 147.3 | -9.237 ± 0.051 | -9.305 ± 0.053 | -10.126 ± 0.047 | 37 | 48.9 |
Credit Screening | 16 | 690 | 416.4 | 204.6 | 69 | -15.26 ± 0.11 | -14.785 ± 0.097 | -17.34 ± 0.11 | 26.9 | 79.9 |
Forest Cover Type | 55 | 28,862 | 23,296 | 2,620 | 2,946 | -16.030 ± 0.050 | -14.455 ± 0.044 | -22.914 ± 0.037 | 288 | 3,629 |
Glass Identification | 10 | 214 | 130.9 | 61.7 | 21.4 | -11.04 ± 0.20 | -11.57 ± 0.17 | -13.201 ± 0.054 | 16.8 | 35 |
Hepatitis | 20 | 155 | 93.2 | 46.3 | 15.5 | -17.68 ± 0.30 | -17.81 ± 0.32 | -18.78 ± 0.37 | 11 | 17.2 |
House Votes | 17 | 435 | 264.4 | 127.1 | 43.5 | -9.90 ± 0.20 | -10.52 ± 0.23 | -14.04 ± 0.19 | 26.1 | 28.2 |
Housing | 14 | 506 | 306.1 | 149.3 | 50.6 | -13.22 ± 0.17 | -13.08 ± 0.16 | -19.497 ± 0.064 | 36.2 | 169.4 |
Image Segmentation | 17 | 2,310 | 1,396 | 651 | 263 | -11.74 ± 0.22 | -11.29 ± 0.21 | -24.4852 ± 0.0043 | 150 | 301 |
Ionosphere | 35 | 351 | 213.7 | 102.2 | 35.1 | -37.49 ± 0.54 | -37.64 ± 0.57 | -52.292 ± 0.051 | 30.7 | 472.4 |
Iris Types | 5 | 150 | 90.8 | 44.2 | 15 | -5.06 ± 0.12 | -5.28 ± 0.12 | -7.375 ± 0.046 | 11.4 | 12.1 |
Isolated Letter Speech | 618 | 7,797 | 5,250 | 988 | 1,559 | -798.7 ± 1.9 | -542.1 ± 1.4 | -912.89 ± 0.28 | 9 | 11,502 |
King Rook vs. King | 7 | 28,056 | 22,640 | 2,548 | 2,868 | -11.217 ± 0.018 | -11.517 ± 0.016 | -13.141 ± 0.019 | 454 | 443 |
Labor Negotiations | 17 | 57 | 25 | 15 | 17 | -21.04 ± 0.87 | -19.93 ± 0.59 | -19.79 ± 0.62 | 11 | 1 |
Landsat | 37 | 6,435 | 3,449 | 986 | 2,000 | -26.70 ± 0.26 | -24.42 ± 0.22 | -59.540 ± 0.030 | 75 | 1,403 |
Letter Recognition | 17 | 20,000 | 16,222 | 1,790 | 1,988 | -15.734 ± 0.098 | -16.48 ± 0.10 | -26.927 ± 0.039 | 666 | 3,950 |
Monks Problem #1 | 7 | 556 | 87 | 37 | 432 | -6.718 ± 0.027 | -6.573 ± 0.019 | -6.7803 ± 0.0094 | 31 | 1 |
Musk | 167 | 476 | 288.9 | 139.5 | 47.6 | -183.4 ± 1.8 | -125.4 ± 1.6 | -261.685 ± 0.089 | 47.8 | 2,592.7 |
New Thyroid | 6 | 215 | 131.5 | 62 | 21.5 | -7.61 ± 0.13 | -7.997 ± 0.078 | -8.832 ± 0.053 | 23 | 4.3 |
Nursery | 9 | 11,025 | 8,972 | 981 | 1,072 | -9.5277 ± 0.0080 | -9.4354 ± 0.0090 | -10.597 ± 0.015 | 177 | 103 |
Page Blocks | 11 | 5,473 | 3,908 | 991 | 574 | -9.24 ± 0.11 | -9.33 ± 0.11 | -16.465 ± 0.041 | 174 | 430 |
Pima Indians Diabetes | 9 | 768 | 466.6 | 224.6 | 76.8 | -11.982 ± 0.055 | -11.843 ± 0.049 | -12.586 ± 0.030 | 29 | 14 |
Poisonous Mushrooms | 23 | 8,124 | 6,351 | 986 | 787 | -9.147 ± 0.022 | -9.222 ± 0.024 | -22.66 ± 0.15 | 233 | 221 |
Promoter | 58 | 106 | 64.3 | 31.1 | 10.6 | -78.81 ± 0.54 | -79.18 ± 0.34 | -79.37 ± 0.34 | 18.8 | 1.4 |
Servo | 5 | 167 | 101.4 | 48.9 | 16.7 | -6.827 ± 0.083 | -6.648 ± 0.073 | -7.695 ± 0.058 | 16.4 | 4.8 |
Shuttle | 10 | 56,793 | 39,157 | 4,343 | 13,293 | -6.958 ± 0.012 | -6.956 ± 0.012 | -11.983 ± 0.012 | 358 | 317 |
Solar Flare | 13 | 1,066 | 641.7 | 317.7 | 106.6 | -5.225 ± 0.058 | -5.319 ± 0.058 | -7.014 ± 0.070 | 15.6 | 23 |
Soybean Large | 36 | 683 | 216 | 91 | 376 | -18.12 ± 0.40 | -17.25 ± 0.35 | -37.3 ± 1.0 | 14 | 145 |
Spambase | 58 | 4,601 | 3,166 | 1,000 | 435 | -13.38 ± 0.21 | -13.53 ± 0.21 | -16.85 ± 0.19 | 36 | 327 |
Splice Junction | 61 | 3,190 | 1,431 | 662 | 1,097 | -79.98 ± 0.38 | -80.01 ± 0.13 | -83.281 ± 0.069 | 717 | 127 |
Thyroid Disease (combined) | 33 | 3,772 | 1,883 | 917 | 972 | -12.97 ± 0.10 | -12.365 ± 0.092 | -16.50 ± 0.12 | 47 | 237 |
Tic-Tac-Toe | 10 | 958 | 578.4 | 283.8 | 95.8 | -9.021 ± 0.020 | -9.644 ± 0.034 | -10.262 ± 0.021 | 35.9 | 77.3 |
Waveform | 22 | 5,000 | 3,495 | 989 | 516 | -29.17 ± 0.15 | -29.49 ± 0.14 | -34.8955 ± 0.0036 | 43 | 150 |
Yeast | 9 | 1,484 | 891.9 | 443.7 | 148.4 | -10.182 ± 0.036 | -10.213 ± 0.035 | -10.870 ± 0.024 | 31 | 30.7 |
Zoo | 17 | 101 | 61.7 | 29.2 | 10.1 | -6.53 ± 0.31 | -7.23 ± 0.33 | -11.77 ± 0.25 | 8.8 | 20.5 |
EachMovie (subset) | 1,648 | 6,117 | 4,524 | 1,002 | 591 | -121.6 ± 5.5 | -120.9 ± 5.6 | -173.5 ± 7.8 | 31 | 5,228 |
Jester | 100 | 17,998 | 14,681 | 1,621 | 1,696 | -95.44 ± 0.73 | -96.29 ± 0.71 | -130.2 ± 1.1 | 142 | 2,552 |
KDD Cup 2000 (subset) | 65 | 13,552 | 9,024 | 976 | 3,552 | -2.103 ± 0.074 | -2.234 ± 0.087 | -2.41 ± 0.11 | 29 | 89 |