An enhancement of random forest algorithm applied in credit card fraud detection system

ABSTRACT One of the most popular frameworks used by data scientists is the random forest algorithm. It is one of the most accurate learning algorithms available. For many data sets, it produces a highly accurate classifier. The random forest algorithm is one of the best among classification algorithms able to classify large amounts of data with accuracy. This study aims to improve the algorithms accuracy by applying our solutions to the problems that always occur in the algorithm. The results should make the algorithms accuracy more accurate in its predictive performance in finding fraudulent transactions inside an e-commerce website single decision trees often have high variance or high bias. Random forest attempts to mitigate the problem of high variance and high bias by engaging to find a natural balance between the attributes that have been used. We have used sampling technique to cut out one third of unnecessary data sets to produce a reliable prediction to our data sets. The results of learning are incomprehensible. Compared to a single decision tree, or to a set of rules, they don't give a lot of insight. Researchers should also improve the tree structure instead of just improving the accuracy itself. Instead of having a big tree structure researchers should also focus on pre-building the tree to select the right attributes on building the tree.

Filipiniana