"Missing is useful": Missing values in cost-sensitive decision trees

Mathematics

• Decision tree 100%
• Missing Values 93%
• Costs 56%
• Misclassification 34%
• Data Mining 14%
• Machine Learning 14%
• Training 13%
• Attribute 12%
• Learning 11%
• Testing 9%
• Strategy 9%
• Cost-sensitive Learning 2%

Engineering & Materials Science

• Decision trees 75%
• Costs 29%
• Cost reduction 13%
• Data mining 10%
• Machine learning 9%
• Testing 6%