# Table 6 Mean number of intervals per predictor variable for the EBD and FI discretization methods

Mean fraction of predictors with 1 interval Mean # of intervals per predictor with >1 interval Mean # of intervals per predictor
Dataset EBD FI EBD FI EBD FI
1 0.81 0.84 2.02 2.01 1.15 1.16
2 0.47 0.61 2.06 2.04 1.48 1.41
3 0.91 0.96 2.02 2.01 1.05 1.04
4 0.18 0.28 2.20 2.16 1.91 1.84
5 0.97 0.99 2.03 2.04 1.01 1.01
6 0.87 0.89 2.01 2.01 1.13 1.11
7 0.82 0.86 2.02 2.01 1.13 1.14
8 0.97 0.98 2.02 2.02 1.01 1.02
9 0.54 0.76 2.11 2.12 1.42 1.27
10 0.38 0.65 2.06 2.06 1.53 1.37
11 0.51 0.77 2.06 2.10 1.41 1.25
12 0.98 0.99 2.02 2.02 1.01 1.01
13 0.05 0.90 2.57 2.10 2.39 1.11
14 0.98 0.99 2.03 2.02 1.01 1.01
15 0.70 0.98 2.08 2.12 1.20 1.02
16 0.75 0.87 2.01 2.01 1.12 1.13
17 0.76 0.85 2.04 2.04 1.16 1.16
18 0.17 0.78 2.31 2.13 1.99 1.25
19 0.87 0.94 2.05 2.02 1.06 1.06
20 0.81 0.85 2.02 2.10 1.15 1.17
21 0.97 0.99 2.01 2.02 1.01 1.01
22 0.82 0.84 2.14 2.14 1.16 1.18
23 0.93 0.97 2.01 2.02 1.05 1.03
24 0.92 0.95 2.06 2.02 1.04 1.05
Average 0.76 0.85 2.08 2.06 1.27 1.16
1. The mean fraction of predictor variables discretized to one interval (no cut points), the mean number of intervals for predictor variables discretized to more than one interval (at least one cut point), and the mean number of intervals for all predictor variables for each dataset is obtained by 10-fold cross-validation done ten times. For each dataset, the higher value is shown in bold font and equal values are underlined.