Đang chuẩn bị nút TẢI XUỐNG, xin hãy chờ
Tải xuống
Chapter 5: Decision trees Introduction to decision tree; Decision tree for pattern recognition; Construction of decision trees; Splitting at the nodes; Overfitting and pruning; Example of decision tree induction. | Chapter 5 Decision trees Assoc. Prof. Dr. Duong Tuan Anh Faculty of Computer Science and Engineering, HCMC Univ. of Technology 3/2015 Outline Introduction to decision tree Decision tree for pattern recognition Construction of decision trees Splitting at the nodes Overfitting and pruning Example of decision tree induction 1. Introduction A decision tree is a tree where each non-leaf node is associated with a decision and the leaf nodes are associated with an outcome or class label. Each internal node test one or more attribute values leading to two or more branches. Each branch in turn is associated with a possible value of the decision. These branches are mutually distinct and collective exhaustive. Decision trees are excellent tools for choosing between several courses of action. They provide a highly effective structure with which you can lay out options and investigate the possible outcome of choosing these options. In the case of binary decision tree, there are two outgoing . | Chapter 5 Decision trees Assoc. Prof. Dr. Duong Tuan Anh Faculty of Computer Science and Engineering, HCMC Univ. of Technology 3/2015 Outline Introduction to decision tree Decision tree for pattern recognition Construction of decision trees Splitting at the nodes Overfitting and pruning Example of decision tree induction 1. Introduction A decision tree is a tree where each non-leaf node is associated with a decision and the leaf nodes are associated with an outcome or class label. Each internal node test one or more attribute values leading to two or more branches. Each branch in turn is associated with a possible value of the decision. These branches are mutually distinct and collective exhaustive. Decision trees are excellent tools for choosing between several courses of action. They provide a highly effective structure with which you can lay out options and investigate the possible outcome of choosing these options. In the case of binary decision tree, there are two outgoing edges from the nodes. One edge represent the outcome “yes” or “true” and the other edge, the outcome “no” or “false”. Decision tree classifier: inductive learning 2. Decision Trees for Pattern Recognition Patterns can be classified using decision trees where the nodes in the tree represent the status of the problem after making some decision. The leaf nodes give the class label of the classification rule corresponding to the path from root node to the leaf node. Example: Consider employee records in a company as in the following table: Name Age Qualification Position ----------------------------------------------------- Ram 55 B. Com. Manager Shyam 30 B. Eng. Manager Mohan 40 M.Sc. Manager Manager No. of Assistants Mood Output ---------------------------------------------------------- Shyam 3 No Medium Shyam 5 No Medium Shyam 1 Yes High Ram 1 Yes Low Ram 5 No Low Ram 5 Yes Low Mohan 1 No Low Mohan 3 Yes Medium Mohan 5 No High Figure 5.1 Decision tree for pattern classification .
