Table 2. Confusion matrices for classification of Twins data.
Predicted | Random forests | Dirichlet multinomial | ||
Actual | Lean | Obese | Lean | Obese |
Lean | 19 | 42 | 33 | 28 |
Obese | 5 | 188 | 29 | 164 |
The two rows give the number of ‘Lean’ and ‘Obese’ individuals predicted to be ‘Lean’ and ‘Obese’ by the random forests and Dirichlet multinomial classifiers following leave-one-out validation. A classification threshold of 0.5 was used for both algorithms.