#TITLE_ALTERNATIVE#

• Main Author: RIZKI HASANAH (NIM: 18214049), NURLAILI
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/29806
• Institution: Institut Teknologi Bandung

<p align="justify"> One of the most popular data mining application in the banking industry and other financial institutions is credit scoring. This technology is very useful and promising because it replaces the old approach that tends to be not objective and depends on the ability of the analyst. One alternative algorithm for data mining that can be applied to credit scoring model development is Artificial Neural Network. In the development process, it is necessary to optimize certain parameters to achieve the best model. This study aims to build credit scoring model using Artificial Neural Network and optimize its parameters. The study used SIJEKH dataset, which is a historical credit data of a financial institution, to be analyzed. The model development process adapted Cross-Industry Standard Process for Data Mining (CRISP-DM), which consists of business understanding, data understanding, data preparation, model development, and evaluation. The modelling step used the popular Scikit-Learn library in Python, MLPClassifier. Optimization is done by selecting the best parameters with one-search at a time and grid search method. Experiments that have been done resulted accuracy, precision, recall, specificity, F-measure, and AUC score of 0.85008658, 0.942447192, 0.86538531, 0.789056875, 0.902259529, and 0.900066602 respectively. Based on the analysis this research, it can be concluded that the most optimal credit scoring model using multi-layer perceptron on Artificial Neural Network is model with ‘Adam’ solver parameter with 607 and 121 number of nodes in the first and second hidden layer, hyperbolic tangent activation function and alpha value of 0.0001. <p align="justify">