Port customer credit risk prediction based on internal and external information fusion

Abstract

Concerning the information asymmetry, subjectivity and other problems in the prediction of port customer credit risk level, a port customer credit risk evaluation system was designed, which provided the quantification base for port customer credit risk prediction. External-internal information fusion model was proposed for port customer credit risk prediction. Customers’ internal history information and relevant external dynamic information were used as information sources. The prediction model fuses external and internal information to predict customers’ credit risk level, therefore assisted port enterprises to estimate customers’ credit risks level more accurately. Top 100 coal industry customers have been selected in experiment, and their operation data were extracted from Guangzhou Port group operation and management system. Relevant external information has also been crawled in experiments correspondingly. Experiment results show that port customer credit risk prediction model based on external-internal information fusion has a better prediction accuracy than non-information-fusion model.