Skip to main content

Table 3 Gradient boosting machine error metrics for patient group membership classification on frequent subgraphs

From: Unsupervised phenotyping of sepsis using nonnegative matrix factorization of temporal trends from a multivariate panel of physiological measurements

Measure Split Subgroup
1 2 3
Precision Train 0.987 0.981 0.988
Test 0.917 0.889 0.939
Recall Train 0.978 0.984 0.990
Test 0.835 0.921 0.951
F-score Train 0.982 0.983 0.989
Test 0.874 0.905 0.945
AUC Train 0.954 0.964 0.977
  Test 0.891 0.914 0.944