<![CDATA[One of problems in lending activities is credit risk due to errors in selecting debtors. In this study, the backpropagation algorithm will be used to develop a prediction calculation system that uses features such as age, gender, marital status, occupation, income, number of dependents, loan amount, time period, collateral, home ownership, and loan purpose to predict creditworthiness. To determine the accuracy level of built, a model evaluation was conducted. The model evaluation was carried out using a confusion matrix, but before that, the data used was separated by ratio of 80 : 20, namely 80% for training and 20% for testing. With the best hyperparameters from several hyperparameter tuning scenarios, the scenario used for implementation in the system is screnario model 5 with 2 hidden layers (50 and 25 neurons), ReLU activation function, learning rate 0.001, 500 epochs, batch size 64, adam optimizer, and early stopping, resulting in an accuracy of 98.18% and a f1 Score of 98.33%. These values are excelent amd show that system created can be used as a reference in predicting creditworthiness. In addition, these values show that the backpropagation model is free from overfitting.]]>
Copyrights © 2025
