8.20 The Fraud Ensemble

8.20 The Fraud Ensemble

For the final phase of this data mining project, we construct an ensemble from the multiple analyses, pooling the results from each algorithm. We partition the data into training, testing, and validating segments, then rotate them using the neural networks, decision tree, and rule generator to identify the fraudulent transactions, mixing and matching the resulting code to optimize the best possible performance from the multiple models. This is a time-consuming process that will require patience and experience, which can only be gained by working through the entire process of knowing the data, preparing it, enhancing it, and analyzing it with multiple algorithms.