Text
PENGARUH RANDOM UNDER-SAMPLING TERHADAP KLASIFIKASI KECURANGAN TRANSAKSI MOBILE MENGGUNAKAN GRADIENT BOOST TREE
Mobile Money Transfer can be defined as a payment transaction process in the form of digital money. This process can make payments, money transfers, and transactions via mobile devices. In the traditional economy toward a digital economy, violations often occur. Control of fraud in the financial sector is not enough to commit fraud crimes. This study builds a mobile transaction classification system that uses the Gradient Boost Tree algorithm. There are problems with the data used, the data is unbalanced. Random undersampling was chosen to solve the data problem. In the application of this algorithm, 10 tests were carried out with the evaluation of k-fold cross-validation on the dataset. From and testing the data, it was found that the average training accuracy is 99.84% and the computation time is 1433.6 seconds for the Gradient Boost Tree algorithm with a parameter value of 100 while the 1000 parameter gets an accuracy of 99.89% with a time of 14044 seconds. In the next scenario, the average accuracy is 97.64% and the computation time is 2.98 seconds for Random Under-sampling and Gradient Boost Trees with a parameter value of 100. The last scenario obtained an average accuracy of 98.57% and computation time of 30.64 seconds for Random Undersampling and Gradient Boost Tree with a parameter value of 1000. There is a decrease inaccuracy due to the Random Under-sampling method that removes data randomly to make the data balanced. Random data deletion causes some important data to be deleted and results in the classification results.
Inventory Code | Barcode | Call Number | Location | Status |
---|---|---|---|---|
2207000974 | T69242 | T692422022 | Central Library (Referens) | Available but not for loan - Not for Loan |
No other version available