Table 2.

Overall linear discriminant analysis (LDA) model performance metrics to separate the four Typha molecular ID classes.

LDA Typha Molecular ID Training Data
Response Variable:	Typha Molecular ID
Predictor Variables:	sqrt[Leaf Count] + sqrt[Longest Leaf Width]
Correctly Predicted:	78.8%
Correctly Predicted 95% CI:	(61.1%, 91.0%)
P-Value [Acc>NIR]:	0.003
Kappa:	66.8%
LDA Typha Molecular ID Permutation Model
Number of Permutations:	1,000
Training / Test Replication:	n = 26 / n = 7
Training Data (% correct ± 1 sd):	87.9% ± 3.6%
Test Data (% correct ± 1 sd):	78.7% ± 15.3%

LDA training data statistics were extracted from a generated confusion matrix in R’s caret package. Permutation model statistics were developed to estimate test data prediction performance for Typha molecular ID classes from both training data and test data. An agreement percentage for predicted data for each iteration was calculated as % correctly predicted cases/total cases predicted and permutation results subsequently averaged.