whiboclustering (version 0.1.2)

whibo_clustering: Find Cluster model using White-Box Cluster Algorithm Design.

Description

Find Cluster model using White-Box Cluster Algorithm Design.

Usage

whibo_clustering(data, k = 3, normalization_type = "No",
  cluster_initialization_type = "Random", assignment_type = "Euclidean",
  recalculation_type = "Mean", max_iteration = 20, no_of_restarts = 1)

Arguments

data

Data on which clustering should be performed.

k

Number of Cluster Representatives.

normalization_type

Which normalization should be used (look at wc_norm_types for possible values). Default value is No.

cluster_initialization_type

Which initialization of Cluster Representatives should be used (look at wc_init_types for possible values). Default value is Random.

assignment_type

Which assignment function should be used (look at wc_assign_types for possible values). Default value is Euclidean.

recalculation_type

Which function for updating Cluster Representatives should be used (look at wc_recalculate_types for possible values). Default value is Mean.

max_iteration

Number of iterations. Default value is 20.

no_of_restarts

Number of restarts of whole clustering procedure. Default value is 1.

Value

Object of type whibo_cluster which include Cluster Representatives (centroids), number of elements per cluster (elements_per_cluster), assignments (assignments), measures of cluster quality (within_sum_of_squares, between_ss_div_total_ss and internal_measures_of_quality), cluster models per iterations (model_history), iterations (iterations) and parameters used (params)

See Also

plot.whibo_cluster, predict.whibo_cluster

Examples

# NOT RUN {
data <- iris[, 1:4] #Take only numerical columns

#Perform k-means clustering
model <- whibo_clustering(data = data, k = 3)
model

#Perform some unorthodox clustering
model <- whibo_clustering(data = data, k = 3,
     normalization_type = 'Z', cluster_initialization_type = 'Ward',
     assignment_type = 'Correlation', recalculation_type = 'Trimean')

# }