Hello everyone.
I'm working with some financial data to predict the failure of the U.S commercial banks throughout 2009 to2014.
One of the models that I'm using is the Dynamic Hazard model.
How can I determine the prediction accuracy of that model. For instance, there is about 500 failed banks in the data set how can I interpret the accuracy of the hazard model to predict the failure ?
Thanks in advance
does the model work on individual banks or banks/ the market in general?
if the model works on individual banks, one would examine its probability of accurately forecasting failures (banks that indeed failed) and its probability of accurately forecasting survivals (banks that survived). also, its likelihood of forecasting a failed bank otherwise, and vice versa. you would have to incorporate and test the model non-failures as well
Hi Ahmed
If you are forecasting failure (dependent variable) with the hazard model the efficacy of your results can be analyzed by the the overall p-value of the model, and so too does each independent variable in the model. If the coefficient of the independent variable is below 1 that variable prevents failure and if the coefficient is above 1 that variable promotes failure. Other metrics to evaluate the overall model are the likelihood ratio chi-ssquared, Harrell's C, and Somers' D.
Raymond A. K. Cox
Thompson Rivers University
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation