Study on Japanese corporate ratings by Rating agencies with Artificial Neural Network and Self-organizing Maps

我们试图验证人工神经网络将成为更强大的工具来模拟只使用数量变量的公司评级，这是由日本的公司机构使用数量变量和定性变量确定的。我们能够明确导致企业评级结果强劲的因素。为了提高预测的准确性，我们期望并尝试使用聚类分析和自组织映射网络。但我们没能得到更好的结果。

We tried to verify that artificial neural network will be stronger tool to simulate only using quantitative variables the corporate ratings that was determined by the corporate agencies using quantitative and qualitative variables in Japan. We are able to make clear the factors that make strong results to the corporate ratings. In order to raise the accuracy, We expected and tried to use the Cluster analysis and self organizing map. But we were not able to get better results.