Decision Tree Optimization in Data Mining with Support and Confidence

Abstract

Decision Tree is a classification technique in data mining that aims to predict behaviour from database. This goal is supported by several algorithms, one of which is Iterative Dichotomiser 3 (ID3) that displays predictions in a tree structure. With the application of decision trees, warehouses or heaps of data can be processed so as to produce rules or decision trees as decision support in solving problems faced by agencies. In fact, the information or rules produced by decision trees are limited to rules using the logic of propositions. The challenge in making decisions on decision trees is how to determine algorithms with a high degree of accuracy from various algorithms in the decision tree and how to find support and confidence for each rule produced by the decision tree to add support value and confidence level of each rule produced. The resulting rule has weaknesses, namely the unavailability of support and confidence, all rules are considered equal in strength based on data before being processed, found records that vary or different amounts of data. By making support and confidence, it will be easier to make decisions based on the results obtained.