Table 1. Simulation parameters representing a scope of clustering problems in clinical research.

Parameters were chosen to represent problems from clinical trials to retrospective electronic health record studies. Single and mixed data types were simulated from all combinations of population characteristics with multiple independent replications.

Population Characteristics	Data Types	Replications
# patients	Single data types	100
200, 800, 3200	Continuous, binary, nominal, ordinal,
# features	categorical1
9, 27, 81, 243	Mixed data types	30
# clusters	Balanced2, unbalanced continuous,
2, 6, 16	unbalanced binary, unbalanced categorical3

¹A mixture of nominal and ordinal data.

²Mixed data simulated with 33% each of binary, categorical, and continuous data.

³Unbalanced mixtures are dominated by 78% of the listed data type.