中文摘要：

我国商业银行信贷记录数据存在历史短、数据多为删失数据等特点，传统统计模型很难对该类数据进行拟合分析。引入医学领域及精算领域使用的生存分析方法，建立个人住房抵押贷款违约风险度量的混合生存模型，并搜集某商业银行历史数据进行实例分析。实例分析的数值模拟表明：所构建模型可以对借款人未来各时点发生违约的概率进行预测和度量，研究结论有助于商业银行准确把握借款人可能违约分布，从而及时有效地管理该类信贷风险。

英文摘要：

The credit data of our commercial banks have a short history. The data are always censored and incomplete. It is hard to fit and to analyze by traditional statistic models. In this paper,we established a mixed survival model based on survival analysis theory which is applied in medical science and actuarial science. Then we made a case study by using the history data of a certain commercial bank. The result shows that the mixed survival model can predict the default time and default probability of the obligors. The conclusion is helpful for commercial bank to predicate the distribution of the default probability and to manage the credit risk.