### Table 2: There are no important class divisions in Britain today

### Table 1. Explanation of the most important classes

### Table 3: Important Words Per Class

"... In PAGE 3: ... We then chose the top words for each of the classes. An example of the ten top words for some of the classes in the veterinary set can be seen in Table3 . Many of the training examples did not contain the important words that were associated with their class.... ..."

### Table 2: Important data for graph classes.

2001

"... In PAGE 13: ... We discovered that b has little impact on the number of crossings reported by any heuristic so it is not mentioned in what follows. Table2 summarizes our experimental subject classes. Within each bigraph type, the size is controlled by the parameter k, called the rank.... In PAGE 16: ... 3.3 Summary We are now ready to take a more detailed look at Table2 . Random classes are denoted R(a; m; d), where d defaults to 1 if it is omitted (random spanning trees for graphs with a gt; 1 are always generated using d = 1, for example).... In PAGE 17: ...TREATMENT EVALUATION 14 Recall that a bigraph type is a collection of classes that di er only in size (number of edges). Table2 shows 23 basic bigraph types | 7 di erent a gt; 1 and 6 di erent d with a = 1 for random graphs; even (square) and odd (rectangular) grids (counted as two distinct types), butter ies, hypercubes5, and 6 types of VLSI graphs. For each type T of 7 isomorphism types, there is a type rnd(T) derived by creating a random graph class with the same m, n0 and n1 as each class in T (there are really 10 isomorphism types, but G00;k has the same parameters as G01;k, etc.... In PAGE 17: ... The total number of classes is 358 (meaning 22; 912 experimental subjects were treated, 64 per class) | 8 sizes for each of 26 classes, the random and their derived isomorphism classes, 10 sizes of grid graphs (times 3 because of the derived random and its derived isomorphism class), 7 of butter ies (times 3), 9 hypercubes (times 3), and 6 size classes for each of 12 VLSI types (6 original, 3 derived random, and 3 derived isomorphism from the random). For all classes listed in Table2 , the average number of crossings reported by tr00, the random placement, is within 0:2% of the expected crossing number m(m ? 1)(n0 ? 1)(n1 ? 1)=4(n0n1 ? 1) reported by War eld [38]. To see the relationship between isomorphism classes and random classes, consider the histograms in Fig.... ..."

Cited by 2

### Table 6. Feature Importance in 3-Classes Using Entropy Criterion

2003

"... In PAGE 10: ... In most cases the results are similar to Multiple Linear Regressions or tree-based software that use statistical methods to measure feature importance. Table6 shows the importance of the six features in the 3-classes case using the En- tropy splitting criterion. Based on entropy, a statistical property called information gain measures how well a given feature separates the training examples in relation to their target classes.... ..."

Cited by 3

### Table 4.10: System overview in MOOSE after import of classes, methods and attributes (Java - third iteration).

### Table 1: Scheduling classes hierarchy ordered by importance.

2003

"... In PAGE 10: ...Table 1: Scheduling classes hierarchy ordered by importance. Table1 lists the scheduling classes we have implemented in our scheduler. FIFO and RR are retained with the same se- mantics as in Linux.... ..."

Cited by 3