Table 6 Paired one-sided DeLong test to compare the AROC values of the four models
From: Predictive models for diabetes mellitus using machine learning techniques
Test name | z-statistic | p-value |
|---|---|---|
GBM vs. Logistic Regression | 1.392 | 0.081 |
GBM vs. Random Forest | 3.885 | 5.13e-05 |
GBM vs. Rpart | 8.914 | 2.20e-16 |
Logistic Regression vs. Random Forest | 2.038 | 0.021 |
Logistic Regression vs. Rpart | 8.006 | 5.95e-16 |
Random Forest vs. Rpart | 7.028 | 1.05e-12 |