Skip to main content
. 2011 May 4;11(5):4899–4916. doi: 10.3390/s110504899

Table 5.

Performances of classification by CART for the two enlarged datasets. Class a), samples non-contaminated; Class b), samples below the legal limit; Class c), samples above the legal limit.

Misclassification matrix Samples fitted assignment) Cross-validated misclassification matrix (Samples predicted assignment)
Analytical protocol True class Total true samples Assigned class Assigned class
a b c a b c
No thermal desorption pre-treatment a 113 112 0 1 111 1 1
Error rate: 0.0082 rate 0.991 0.000 0.009 0.982 0.009 0.009
Risk: 0.0082 b 6 0 6 0 0 5 1
Cross-validated Error Rate: 0.0328 rate 0.000 1.000 0.000 0.000 0.833 0.167
Cross-validated Risk: 0.0328 c 3 0 0 3 0 1 2
rate 0.000 0.000 1.000 0.000 0.333 0.667
  Desorption temperature 180 °C a 113 112 1 0 110 1 2
Error rate: 0.0328 rate 0.991 0.009 0.000 0.973 0.009 0.018
Risk: 0.0328 b 6 3 3 0 5 1 0
Cross-validated Error Rate: 0.0656 rate 0.500 0.500 0.000 0.833 0.167 0.000
Cross-validated Risk: 0.0656 c 3 0 0 3 0 0 3
rate 0.000 0.000 1.000 0.000 0.000 1.000