| dc.description.abstract |
PT Permodalan Nasional Madani (PNM) Simalungun branch is a financial institution that provides loan (credit) services to the community. In the activity of distributing loans, there is the term credit risk that must be managed. This credit risk is the failure or inability of the borrower to fulfill its obligations in terms of making payments based on previously agreed terms. In credit risk, late loan payments are the main problem in credit risk which can cause detrimental financial impacts and disrupt business operations. Therefore, to overcome this problem, PNM Mekaar Simalungun branch can take advantage of technological advances in the field of artificial intelligence (Artificial Intelligence). This research aims to overcome these challenges by building an Artificial Neural Network (ANN) model using the Backpropagation algorithm to predict delays in loan payments at the Simalungun branch of PNM Mekaar based on related variables or predetermined features. The ANN model with the Backpropagation algorithm that was built has optimal performance. The model is built with a layer arrangement (hidden layer1=80 with RELu activation and input layer=7, hidden layer2=8 with dropout=0.5 and RELu activation, hidden layer3=8 with dropout=0.3 and RELu activation, hidden layer4= 8 with dropout=0.1 and RELu activation, hidden layer5=1 with Sigmoid activation). The model testing results after being evaluated using testing data, obtained an accuracy score of 99.1%, a precision score of 100%, a recall score of 98.9%, and an f1-score of 99.4%. The model was trained using parameters batch size 64, optimizer RMSprop, learning rate 0.001, and number of epochs 100.
Keywords : Prediction, Machine Learning, Artificial Neural Network, Backpropagation
| en_US |
