DataMining.ClusterAnalysis.Iris
abstract class DataMining.ClusterAnalysis.Iris
The test and demo for application of Cluster Analysis to Iris dataset contained in class DataMining.IrisDataset
Run method ##class(DataMining.ClusterAnalysis.Iris).test()
See also Graphical Zen Demo implemented in class DataMining.ClusterAnalysis.ZenDemoPage
Method Inventory
Methods
classmethod DeleteModel() as %Status
classmethod GetModel(method As %String, Output sc As %Status) as %DeepSee.extensions.clusters.AbstractModel
classmethod GetResults(model As %DeepSee.extensions.clusters.AbstractModel, Output correct As %Integer, Output error As %Integer, Output s1 As %String, Output s2 As %String, Output s3 As %String)
classmethod Summary(model As %DeepSee.extensions.clusters.AbstractModel, Output pClusterNames) as %Status
classmethod checkData() as %Status
classmethod speciesMatrix(model As %DeepSee.extensions.clusters.AbstractModel, k As %Integer, ByRef matrix, Output mostFrequent As %String) as %Status
This method runs the full Cluster Analysis procedure for Iris dataset.
Parameters:
- verbose - Level of verbosity. Defualt is 2 that prints the information readable on a single screen. Increasing verbosity might yield to too much information being printed but provides more details. Verbosity 1 prints just the results but no processing information. Set to 0 for silent run.
- NClusters - Number of clusters. The default is 3 and corresponds to the number of actual species in the dataset.
- method - Algorithm to use. Default is 1 which will run "PAM with SA" implemented by class %DeepSee.extensions.clusters.PAMSA. Setting to 0 will run simple and pure PAM algorithm implemented in %DeepSee.extensions.clusters.PAM which is faster but produces less reliable results. Setting it to 2 will run CLARA implemented by %DeepSee.extensions.clusters.CLARA which is certainly an overkill for a small dataset like Iris.