Complexity Analysis of Consumer Finance following Computer LightGBM Algorithm under Industrial Economy

Abstract

In the advancement of communication technologies and electronic commerce, the industrial economy consumer finance serves as the source of financial stability and improves the economic and social status of the household; thus, there is a need to significantly prevent default in consumer finance. The prediction of individual default and prevent default in consumer finance has become a significant factor promoting the growth of the industrial economy in the financial sector. Thus, there is a need for an effective and efficient approach for promoting the industrial economy. This study aims to improve the prediction accuracy of individual default and prevent default in consumer finance using an optimized light gradient boosting machine (LightGBM) algorithm. The principles of LightGBM are explored, and the key factors affecting the performance of LightGBM are analyzed. The prediction performance of LightGBM is improved by balancing the training dataset. The performance of LightGBM is compared with several machine learning algorithms using Alibaba Cloud Tianchi big datasets. The experimental results show that the LightGBM prediction model achieved the highest performance with an accuracy of 81%, precision 88%, recall 72%, the area under the curve (AUC) with 0.76, and the F1 score (F1) with 0.79. The optimization of LightGBM can greatly enhance the prediction of personal default, which is helpful to the effective analysis of consumer finance complexity, reducing the investment risk of the financial industry and promoting the development of the industrial economy in the financial sector.