Predicting Corporate Credit Ratings Using Content Analysis of Annual Reports – A Naïve Bayesian Network Approach

摘要: Corporate credit ratings are based on a variety of information, including financial statements, annual reports, management interviews, etc. Financial indicators critical to evaluate corporate creditworthiness. However, little is known about how qualitative information hidden in firm-related documents manifests rating process. To address this issue, study aims develop methodology for extracting topical content from using latent semantic analysis. This integrated with traditional into multi-class prediction model. Informative obtained correlation-based filter the process feature selection. We demonstrate that Naive Bayesian networks perform statistically equivalent other machine learning methods terms classification performance. further show “red flag” values may indicate low quality (non-investment classes) firms. These findings can be particularly important investors, banks and market regulators.