Certainty factor based pruning for C4.5 algorithm

Classification is the process of group data into data classes according to their patterns. One method that can classify and show relationships between attributes is the decision tree method. The classification method using this decision tree is still challenging to improve, especially those related to increasing accuracy. One factor that can reduce accuracy is the pruning process of the branches (nodes) with high information values. Thus, this study aims to optimize the pruning process. The pruning process unexpectedly removes some contributive nodes (nodes with high information). The pruning method uses a certainty factor (CF) introduced to ensure that the trimmed nodes are non-contributing nodes. Improving the performance of the C4.5 algorithm using CF has succeeded in minimizing the level of errors caused by the pruning process.