Table 1.

Summary of the quality metrics of the unsupervised machine learning models centroid k-means and medoid-PAM to distinguish root system architecture in alfalfa. B is for branch rooted type, T is for tap rooted, and TB is intermediate. μ is the mean for each metric.

Metric	Model 1: centroid k-means				Model 2: medoid-PAM
	B	T	TB	μ	B	T	TB	μ
Sensitivity	0.828	0.763	0.305	0.632	0.738	0.658	0.229	0.541
Specificity	0.732	0.771	0.860	0.788	0.740	0.805	0.786	0.777
Positive predictive value	0.447	0.604	0.667	0.573	0.498	0.722	0.319	0.513
Negative predictive value	0.942	0.876	0.575	0.798	0.889	0.753	0.699	0.780
Precision	0.447	0.604	0.667	0.573	0.498	0.722	0.319	0.513
Prevalence	0.207	0.314	0.478	0.333	0.259	0.436	0.305	0.333
Balanced accuracy	0.780	0.767	0.583	0.710	0.739	0.731	0.507	0.659