A novel corporate credit rating system based on Student’s-t hidden Markov models

作者：

Highlights：

• We propose a credit rating system based on hidden Markov models.

• Our system captures strong temporal dynamics in the data.

• Robust to outliers in the training data.

• State-of-the-art results in corporate default prediction.

摘要

•We propose a credit rating system based on hidden Markov models.•Our system captures strong temporal dynamics in the data.•Robust to outliers in the training data.•State-of-the-art results in corporate default prediction.

论文关键词：Corporate credit rating,Hidden Markov model,Student’s-t distribution,Expectation-maximization,Basel framework,Statistical machine learning

论文评审过程：Received 2 December 2014, Revised 8 January 2016, Accepted 9 January 2016, Available online 28 January 2016, Version of Record 11 February 2016.

论文官网地址：https://doi.org/10.1016/j.eswa.2016.01.015