Forward Intensity Model Monitoring Using Multivariate Exponential Weighted Moving Average Scheme

Authors

Linzi Liu, Xi'an Jiaotong UniversityFollow
Xin Lai, Xi'an Jiaotong UniversityFollow
Yuting Rong, Xi'an Jiaotong UniversityFollow
Wei Huang, Xi'an Jiaotong UniversityFollow

Abstract

We propose a parameter monitoring method for the forward intensity model – the default probability prediction model of the Credit Research Initiative (CRI). We review the relative statistical process control scheme in the field of engineering. Based on this, we propose a new Multivariate Exponential Weighted Moving Average (MEWMA) scheme to monitor the forward intensity model monthly. This new chart might be applied to identify and diagnose the out-of-control (OC) parameters in real time as the data updating, which reduces the cost of recalculating all parameters and improve the operational and calculational efficiency of the default prediction models in practical application.

Recommended Citation

Liu, Linzi; Lai, Xin; Rong, Yuting; and Huang, Wei, "Forward Intensity Model Monitoring Using Multivariate Exponential Weighted Moving Average Scheme" (2019). PACIS 2019 Proceedings. 99.
https://aisel.aisnet.org/pacis2019/99