Applying A Machine Intelligence Algorithm for Prediction

This paper presents an application of the C4.5 algorithm for generating prediction rules in various real-world data sets. The data sets consist of different kinds of attribute types including numerical, categorical, and mix types. In our experiments, the C4.5 pruned/unpruned tree algorithm is applied to four different kinds of data sets obtained from the UCI Machine Learning Repository and pools of wells in the region of Saskatchewan, Canada. The result showed that the C4.5 algorithm performed well on categorical, numerical and mix-type data. The pruned C4.5 tree algorithm demonstrated having better performance than the unpruned one when these types of data are concerned. It also discusses the limitations of this algorithm and possible extension when predicting future production performance of oil wells