2022 Volume 32 Issue 4 Pages 133-154
Abstract. This study examines the use of financial time series data as input data for the corporate credit rating classification problem. Specifically, we employed a long short-term memory model (LSTM), which has a structure suitable for handling time series data, for the classification problem on Japanese credit rating data, and verified whether the accuracy of the classification is improved or not. As a result, we observed that the longer the time series length of the input financial data, the higher the classification accuracy for the test data set.