Clustering of Bangladeshi female immigrants in West Bengal

Pranati Datta, Indian Statistical Institute

Cluster analysis is typically used to identify relatively homogeneous subgroups within a more heterogeneous population. This study is devoted to find out whether useful grouping or cluster exists among Bangladeshi female immigrants in West Bengal, a state in India. Volume of female immigrants from Bangladesh by districts of West Bengal, an Indian state, has been obtained from Census of India , 2001. Agglomerative hierarchical average limkage within group clustering techniques have been used in the present study to identify homogeneous sub groups of Bangladeshi immigrants in districts of West Bengal. Clustering is grouping of objects that are homogeneous in terms of standard similarity judgement. Similarity judgement may be made in terms of a similarity measure or a distance measure. In our present data we have used special case of Minkowski distance measure. After preparing similarity matrix average linkage(within group) clustering has been applied to similarity matrix based on our univariate data. It joins the two clusters for which the average distance between members of the resulting cluster will be smallest to create homogeneity within groups . Proximity matrix, agglomeration schedule and dendrogram have been prepared to identify different clusters. Dendrograms, also called hierarchical tree diagrams, show the relative size of the proximity coefficients at which districts are combined . The first cluster is formed by districts Bankura and Birbhum at .060 distance coefficient. Darjiling, Murshidabad, Hugli, Bardhaman, Maldah, South 24 Parganas, Malda, Medinipur, Darjiling, Howrah, and other districts are merged gradually and different sub-clusters are formed at different distance levels. Districts Nadia and North 24 Parganas are merged at distance coefficients 4.867 and 7.345 respectively. These two districts are mostly dissimilar member and hence they are combined at highest distance level. This dissimilarity is supported by the fact that North 24 parganas receives highest Bangladeshi immigrants followed by districts Nadia.

  See paper

Presented in Poster Session 2