Implementation of Adaptive Neuro-Fuzzy Inference System (ANFIS) Algorithm for Customer Credit Prediction

Credit policies in the sales process are one of the supplier's most powerful tools for controlling demand, increasing sales and promoting commodities. Any supply chain member wishing to buy will receive full or partial trade credit based on strong decision-making rights. The problem that often occurs is the large number of bad loans that can disrupt the economic cycle. This problem has also resulted in many finance companies experiencing difficulties in processing customer loans. This problem can be prevented by applying artificial intelligence to predict which customers will take credit. Predictions will be made using the ANFIS method. Adaptive neuro-fuzzy inference system (ANFIS) methods deal with uncertainty in predictions and have been widely applied in various research contexts including medical systems, image processing, electrical systems, and so on. We chose this methodology because it has mostly proven to perform well for modeling complex data sets. The study used 200 customer data taken from a finance company, the data collected is customer data that can pay on time and customer data that cannot pay on time. The main novelty of this research lies in the application of a hybrid model that combines neural networks and fuzzy inference systems to predict credit risk. Most existing models use standard techniques such as logistic regression or decision tree-based models. By using ANFIS, which combines the ability of neural networks to handle non-linear relationships and fuzzy logic to model uncertainty, this research brings a new, more complex and more accurate approach to credit assessment. The results obtained from this study were as much as 80.5% of the data according to or valid with the predictions made using the ANFIS method. This proves that the model applied to the ANFIS method can be used to predict customer credit in the future.