Cluster analysis
Based on H.C. Romesburg: Cluster analysis for researchers, Lifetime Learning Publications, Belmont, CA, 1984 P.H.A. Sneath and R.R. Sokal: Numericxal Taxonomy, Freeman, San Fransisco, CA, 1973
Jens C. Frisvad BioCentrum-DTU
Biological data analysis and chemometrics
Two primary methods
Cluster analysis (no projection)
Hierarchical clustering Divisive clustering Fuzzy clustering
Ordination (projection)
Principal component analysis Correspondence analysis Multidimensional scaling
Advantages of cluster analysis
Good for a quick overview of data Good if there are many groups in data Good if unusual similarity measures are needed Can be added on ordination plots (often as a minimum spanning tree, however) Good for the nearest neighbours, ordination better for the deeper relationships
Different clustering methods
NCLAS: Agglomerative clustering by distance optimization HMCL: Agglomerative clustering by homogeneity optimization INFCL: Agglomerative clustering by information theory criteria MINGFC: Agglomerative clustering by global optimization ASSIN: Divisive monothetic clustering PARREL: Partitioning by global optimization FCM: Fuzzy c-means clustering MINSPAN: Minimum spanning tree REBLOCK: Block clustering (k-means clustering)
SAHN clustering
Sequential agglomerative hierarchic nonoverlapping clustering
Single linkage
Nearest neighbor, minimum method Close to minimum spanning tree Contracting space Chaining possible J = 0.5, K = 0.5, = 0, = -0.5 UJ,K = min Ujk
U( J ,K ) L = JUJ ,L + KUK ,L + UJ ,K + UJ ,L UK ,L
Complete linkage
Furthest neighbor, maximum method Dilating space J = 0.5, K = 0.5, = 0, = 0.5 UJ,K = max Ujk
Average linkage
Aritmetic average
Unweighted: UPGMA (group average) Weighted: WPGMA
Centroid
Unweighted centroid (Centroid) Weighted centroid (Median)
From Sneath and Sokal, 1973, Numerical taxonomy
Ordinary clustering
Obtain the data matrix Transform or standardize the data matrix Select the best resemblance or distance measure Compute the resemblance matrix Execute the clustering method (often UPGMA = average linkage) Rearrange the data and resemblance matrices Compute the cophenetic correlation coefficient
Binary similarity coefficients
(between two objects i and j)
j i 1
Matches and mismatches
m = a + b (number of matches) u = c + d (number of mismatches) n = m + u = a + b + c + d (total sample size) Similarity (often 0 to 1) Dissimilarity (distance) (often 0 to 1) Correlation (-1 to 1)
Simple matching coefficient
SM = (a + d) / (a + b + c + d) = m / n Euclidean distance for binary data: D = 1-SM = (b +c) / (a + b + c + d) = u / n
Avoiding zero zero comparisons
Jaccard = J = a / (a +b +c) Srensen or Dice: DICE = 2a / (2a + b + c)
Correlation coefficients
Yule: (ad bc) / (ad + bc)
PHI = ( ad bc ) / ( a + b )(c + d )( a + c )(b + d )
Other binary coefficients
Hamann = H = (a + d b c) / (a + b + c + d) Rogers and Tanimoto = RT = (a + d) / (a + 2b + 2c + d) Russel and Rao = RR = a / (a + b + c + d) Kulzynski 1 = K1 = a / (b + c) UN1 = (2a + 2d) / (2a + b + c + 2d) UN2 = a / (a + 2b + 2c) UN3 = (a + d) / (b + c)
Distances for quantitative (interval) data Euclidean and taxonomic distance
EUCLID = Eij =
DIST = d ij = 1 n
( xki + xkj )
( x ki + x kj ) 2 k
Bray-Curtis and Canberra distance
BRAYCURT = d ij = k xki xkj / k ( x ki + xkj )
1 CANBERRA = k xki xkj / k ( xki + xkj ) n
Average Manhattan distance (city block)
1 MANHAT = M ij = k xki xkj n
Chi-squared distance
xki xkj x. x. j i CHISQ = d ij = k xk
Cosine coefficient
COSINE = cij = k xki xkj /
k xki
k xkj
Step 1. Obtain the data matrix
Object 1 1 Feature 2 2 3 4 5
10 5
20 20
30 10
30 15
5 10
Objects and features
The five objects are plots of farm land The features are
1. Water-holding capacity (%) 2. Weight % soil organic matter
Objective: find the two most similar plots
Resemblance matrix 1 1 2 18.0 3 4 2 3 4 5
14.1 11.2 18.0
5.00 25.0
25.5
20.6 22.4
5 7.07
Revised resemblance matrix 1 1 2 5 (34) 2 5 (34)
18.0 7.07 21.5
18.0 12.7
25.3
Revised resemblance matrix
2 2
(34)
(15)
12.7 18.0
23.4
(34)
(15)
Rvised resemblance matrix (15) (234)
(15)
21.6
(234)
Cophenetic correlation coefficient (Pearson product-moment correlation coefficient)
A comparison of the similarities according to the similarity matrix and the similarities according to the dendrogram
rX ,Y =
xy (1/ n)(x)(y) (x (1/ n)(x) )(y (1/ n)( y) )
2 2 2 2
NTSYS
Import matrix Transpose matrix if objects are rows (they are supposed to be columns in NTSYS) (transp in transformation / general) Consider log1 or autoscaling (standardization) Select similarity or distance measure (similarity) Produce similarity matrix
NTSYS (continued)
Select clustering procedure (often UPGMA) (clustering) Calculate cophenetic matrix (clustering) Compare similarity matrix with cophenetic matix (made from the dendrogram) and write down the cophenetic correlation (graphics, matrix comparison) Write dendrogram (graphics, treeplot)
