XGBoost-Driven Harsanyi Transformation and Its Application in Incomplete Information Internet Loan Credit Game

  • Yi-Cheng GongEmail author
  • Yan-Na Zhang
  • Li Yu
Conference paper
Part of the Communications in Computer and Information Science book series (CCIS, volume 1082)


In theory, the key step of traditional Harsanyi transformation is “Nature” assigns types to real players according to certain probability distributions. In big data era, it is still a challenge to obtain the probability distributions in practice, partly because the history type data of a new player is still private. Considering it is easy to access some feature data of a new player as well as to obtain both the feature and type data of massive other players, this paper introduces the statistical learning method eXtreme Gradient Boosting (XGBoost) to propose an XGBoost-driven Harsanyi transformation, where XGBoost is used to predict a new player’s type distribution indirectly. To test the effect of XGBoost-driven Harsanyi transformation, an incomplete information Internet loan credit game (3ILCG) is modeled and analyzed. When the loan interest rate r = 0.2, the empirical analysis is executed on 24,000 training data and 6,000 testing data. The experiment shows the accuracy (A) and harmonic mean (F1) of the enterprise loan decision based on pxgb on 6,000 testing data are 0.900833 and 0.945864 respectively. The test experiment demonstrates the XGBoost-driven Harsanyi transformation can help the lending platform to make loan decisions scientifically in practice and improve the practice value of game theory.


Game theory Bayesian Nash equilibrium Harsanyi transformation XGBoost Internet loan 



Thanks to the support from National Natural Science Foundation of China (Project Number: 61671338).


  1. 1.
    Harsanyi, J.C.: Games with incomplete information played by “Bayesian’’ players, I-III Part I, the basic model. Manag. Sci. INFORMS 14(3), 159–182 (1967)zbMATHGoogle Scholar
  2. 2.
    Harsanyi, J.C., Selten, R.: A generalized Nash solution for two-person bargaining games with incomplete information. Manag. Sci. 18, 80–106 (1972)MathSciNetzbMATHCrossRefGoogle Scholar
  3. 3.
    Harsanyi, J.C.: Games with randomly disturbed payoffs: a new rationale for mixed-strategy equilibrium points. Int. J. Game Theory 2(1), 1–23 (1973)MathSciNetzbMATHCrossRefGoogle Scholar
  4. 4.
    Xiong, F., Liu, Y., Si, X., Chen, H.: Group decision-making simulation with incomplete information. Syst. Eng. Theory Pract. 31(1), 151–157 (2011)Google Scholar
  5. 5.
    Huang, H., Sun, Y.E., Chen, Z.L., Xu, H.L., Xing, K., Chen, G.L.: A study on complete competitive equilibrium of spectrum two-way auction mechanism. Comput. Res. Dev. 51(3), 479–490 (2014)Google Scholar
  6. 6.
    Yang, Y., Zhu, W., Zhang, X.: New mechanism design in the C2C online reputation evaluation optimizing. In: Proceedings of the 13th International Conference on Enterprise Information Systems, pp. 1021–1029. Springer, Berlin (2015)Google Scholar
  7. 7.
    Shun, Z., Mingshun, L.I.: Game analysis of PPP project risk sharing under the condition of incomplete information. Eng. Econ. 27(4), 37–41 (2017)Google Scholar
  8. 8.
    Gong, Y., Zhang, Y., Zhao, J., Yu, L. Pan, T.: Incomplete information game and simulation of logistics supervision. In: 2017 2nd International Conference on Applied Mathematics, Simulation and Modeling, pp. 284–289. Atlantis Press, Bangkok (2017)Google Scholar
  9. 9.
    He, D., Chen, W., Wang, L., Liu, T.Y.: A Game-theoretic machine learning approach for revenue maximization in, sponsored search. In: Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence, pp. 284–289. The AAAI Press, Palo Alto (2013)Google Scholar
  10. 10.
    Xu, H., Gao, B., Yang, D., Liu, T.Y.: Predicting advertiser bidding behaviors in sponsored search by rationality modeling. In: Proceedings of the 22nd International conference on World Wide Web, pp. 176–186. Association for Computing Machinery Press, New York (2013)Google Scholar
  11. 11.
    Li, H., Tian, F., Chen, W., Qin, T., Liu, T. Y: Generalization analysis for game- theoretic machine learning. In: Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence, pp. 2089–2095. The AAAI Press, Palo Alto (2014)Google Scholar
  12. 12.
    Liu, T.Y., Chen, W., Qin, T.: Mechanism learning with mechanism induced data. In: Twenty-ninth AAAI Conference on Artificial Intelligence, pp. 4037–4041. The AAAI Press, Palo Alto (2015)Google Scholar
  13. 13.
    Laird, J.E.: Research in human-level AI using computer games. Commun. ACM 45(1), 32–35 (2002)CrossRefGoogle Scholar
  14. 14.
    Spronck, P., Ponsen, M., Sprinkhuizen-Kuyper, I., Postma, E.: Adaptive game AI with dynamic scripting. Mach. Learn. 63(3), 217–248 (2006)CrossRefGoogle Scholar
  15. 15.
    Gibney, E.: Self-taught AI is best yet at strategy game Go. Nature 10(1), 68–74 (2017)MathSciNetGoogle Scholar
  16. 16.
    Silver, D., Huang, A., Maddison, C.J., Guez, A., Hassabis, D.: Mastering the game of go with deep neural networks and tree search. Nature 529(7587), 484–489 (2016)CrossRefGoogle Scholar
  17. 17.
    Elizabeth, G.: Google AI algorithm masters ancient game of Go. Nature 529(7587), 445–446 (2016)CrossRefGoogle Scholar
  18. 18.
    Xie, Z.: Economic Game Theory, 4th edn. Fudan University Press, Shanghai (2002)Google Scholar
  19. 19.
    Chen, T., Guestrin, C.: Xgboost: a scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery Press, New York (2016)Google Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2019

Authors and Affiliations

  1. 1.Hubei Province Key Laboratory of Systems Science in Metallurgical ProcessWuhan University of Science and TechnologyWuhanChina
  2. 2.Department of Mathematics and Statistics, Science CollegeWuhan University of Science and TechnologyWuhanChina

Personalised recommendations