2.3 Model Results and Robustness

The Concordance Index, often referred to as the c-index or C-statistic, is a measure of predictive accuracy commonly used in survival analysis and time-to-event modelling.

In survival analysis, the c-index is calculated by comparing pairs of individuals in the dataset, and evaluating whether the model correctly ranks the predicted survival times. The c-index is then computed as the ratio of concordant pairs to the total number of non-tied pairs (concordant + discordant).

• Concordant pairs: Pairs of observations where the model correctly predicts the order of survival times.

• Discordant pairs: Pairs of observations where the model incorrectly predicts the order of survival times.

The c-index ranges from 0 to 1, where:

• A c-index of 0.5 indicates random chance (no predictive power).

• A c-index of 1.0 indicates perfect predictive accuracy.

Concordance= 0.8711 with se= 0.01962 for Corporate debt

Concordance= 0.8044 with se= 0.02081 for Project debt