Improving Predictions of Multiple Binary Models in ILP

Table 1

Coverage of three rules over a problem of 15 examples 𝑥₁,…,𝑥₁₅ belonging to three classes 𝜆₁,…,𝜆₃.

Rule	Class predicted	Coverage	𝜒^2 score
𝑟1	𝜆1	𝑥1,…,𝑥5,𝑥9,𝑥13	8.57
𝑟2	𝜆3	𝑥5,…,𝑥9,𝑥14,𝑥15	3.75
𝑟3	𝜆3	𝑥9,𝑥11,…,𝑥15	11.67
Uncovered	𝜆2; (Majority of uncovered)	𝑥10	N.A.