DaveM@9: function linkList = aglomCluster(data, clusterMethod, distanceMetric, numClusters)
DaveM@8: %% aglomCluster(data, clusterMethod, distanceMetric, numClusters)
DaveM@8: % This function performs aglomerative clustering on a given data set,
DaveM@8: % allowing the interpretation of a hierarchical data, and plotting a
DaveM@8: % dendrogram.
DaveM@8: %
DaveM@8: % data in the format of of each row is an observation and each column is a
DaveM@8: % feature vector clusterMethod;
DaveM@8: %     * 'average'     Unweighted average distance (UPGMA)
DaveM@8: %     * 'centroid'	Centroid distance (UPGMC), appropriate for Euclidean
DaveM@8: %     distances only
DaveM@8: %     * 'complete'	Furthest distance
DaveM@8: %     * 'median'      Weighted center of mass distance (WPGMC),appropriate
DaveM@8: %     for Euclidean distances only
DaveM@8: %     * 'single'      Shortest distance
DaveM@8: %     * 'ward'        Inner squared distance (minimum variance algorithm),
DaveM@8: %     appropriate for Euclidean distances only (default)
DaveM@8: %     * 'weighted'	Weighted average distance (WPGMA)
DaveM@8: % distanceMetric
DaveM@8: %     * 'euclidean' Euclidean distance (default).
DaveM@8: %     * 'seuclidean' Standardized Euclidean distance. Each coordinate
DaveM@8: %     difference between rows in X is scaled by dividing by the
DaveM@8: %     corresponding element of the standard deviation S=nanstd(X). To
DaveM@8: %     specify another value for S, use D=pdist(X,'seuclidean',S).
DaveM@8: %     * 'cityblock' City block metric.
DaveM@8: %     * 'minkowski' Minkowski distance. The default exponent is 2. To
DaveM@8: %     specify a different exponent, use D = pdist(X,'minkowski',P), where P
DaveM@8: %     is a scalar positive value of the exponent.
DaveM@8: %     * 'chebychev' Chebychev distance (maximum coordinate difference).
DaveM@8: %     * 'mahalanobis'	Mahalanobis distance, using the sample covariance
DaveM@8: %     of X as computed by nancov. To compute the distance with a different
DaveM@8: %     covariance, use D = pdist(X,'mahalanobis',C), where the matrix C is
DaveM@8: %     symmetric and positive definite.
DaveM@8: %     * 'cosine' One minus the cosine of the included angle between points
DaveM@8: %     (treated as vectors).
DaveM@8: %     * 'correlation' One minus the sample correlation between points
DaveM@8: %     (treated as sequences of values).
DaveM@8: %     * 'spearman' One minus the sample Spearman's rank correlation between
DaveM@8: %     observations (treated as sequences of values).
DaveM@8: %     * 'hamming' Hamming distance, which is the percentage of coordinates
DaveM@8: %     that differ.
DaveM@8: %     * 'jaccard' One minus the Jaccard coefficient, which is the
DaveM@8: %     percentage of nonzero coordinates that differ.
DaveM@8: % numClusters is the number of final clusters produced by the dendrogram,
DaveM@8: % if 0 (default), then will infer from data
DaveM@8: 
DaveM@8: if(nargin<2)
DaveM@8:     clusterMethod = 'ward';
DaveM@8: end
DaveM@8: if(nargin<3)
DaveM@8:     distanceMetric = 'euclidean';
DaveM@8: end
DaveM@8: if (nargin<4)
DaveM@8:     numClusters = 0;
DaveM@8: end
DaveM@8: 
DaveM@8: distMap = pdist(data, distanceMetric);
DaveM@8: linkList = linkage(distMap, clusterMethod);
DaveM@36: [~,T] = dendrogram(linkList,numClusters,'Orientation','left');
DaveM@8: 
DaveM@8: 
DaveM@8: end