boostKCP : Boosting K-means Clustering for the Pearson correlation distance (last-modified: 18th March 2013)

Note: One has to compile the boostKCP.cpp file before running the program according to each operating system.

	e.g. g++ -std=C11++ -o boostKCP boostKCP.cpp

Usage:
boostKCP input_data_name number_of_clusters number_of_trials (-O output_data_name_prefix) (-B) (-C) (-L) (-M) (-N)

	EXAMPLE
	boostKCP input_file.txt 10 5 -O output_file -B -C -L -M

REQUIRED
	input_data_name <filename of input> 
	number_of_clusters<number of clusters in K clustering>
	number_of_trials<number of trials of K clustering>

OPTIONAL
	-O output_data_name_prefix <prefix of output>
		Default is input_data_name without extersion
	-B <bound B mode>
		execute K clustering with Pruning method of bound B; default is Pruning method of bound A
	-L <Lloyd Algorithm mode>
		execute K clustering without Pruning method; default is Pruning method of bound A
	-C <compare mode>
		execute K clustering with Pruning methods and without Pruning method with same first k centroids
	-M <K-median clustering mode>
		execute K-median clustering; default is K-means clustering

INPUT FORMAT
	actual value data divided by space character or tab character is available
	each line are read into each data point
	dimension of each data must be same length

OUTPUT FORMAT
	g(output_data_name_prefix)_time_(name of mode).txth
		total calculation time of each trial
	g(output_data_name_prefix)_vector_(name of mode)_(number of trial).txth
		values of centroid vectors in each iteration of each trials
	g(output_data_name_prefix)_cluster_(name of mode)_(number of trial).txth
		list of clustered data points of each trial

CAUTION
	Pearson correlation distance can not be calculated between vectors which variance are zero
	Input vectors which variance are zero are automatically removed