Abstract
American Express (AMEX) is one of the most popular credit card services in the U.S. Lately, the company is facing loss of some valued customers for specific reasons. This paper investigates the customers’ cancellation behavior based on XGBoost in terms of the AMEX data. Besides, the factors contributing to customers’ cancellations are identified, e.g., customer age, total transaction counts, total transaction amount. According to the analysis, the proposed XGBoost model can reach 96.5% in accuracy. By implementation of such tool, it is feasible for the bank to predict their customer behavior and take measures proactively. These results shed light on guiding further exploration of credit card services.
Reference12 articles.
1. The Anticompetitive Effects of Vertical Most-Favored-Nation Restraints and the Error of Amex
2. M. Dumiak, Advertising Campaigns: Amex Unrivaled in Advertising Spending: The three top spenders in 1999 were credit card companies, but it was also the year of the dot-com effect, Financ. Serv. Mark., vol. 2, no. 4, p. 8, 2000.
3. Sample size requirements for estimating pearson, kendall and spearman correlations
4. Temporospatial variations and Spearman correlation analysis of ozone concentrations to nitrogen dioxide, sulfur dioxide, particulate matters and carbon monoxide in ambient air, China
5. J. O. May and S. W. Looney, “Sample size charts for Spearman and Kendall coefficients,” J. Biom. Biostat., vol. 11, no. 6, pp. 1–7, 2020.