Date of Submission

2-28-1988

Date of Award

2-28-1989

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Doctoral Thesis

Degree Name

Doctor of Philosophy

Subject Name

Computer Science

Department

Electronics and Communication Sciences Unit (ECSU-Kolkata)

Supervisor

Majumdar, Dwijesh Dutta (ECSU-Kolkata; ISI)

Abstract (Summary of the Work)

Brief reviow of oluater analysis. The literature on oluater analy ste is baaically orianted touerda the development of algori thme 1,2,3, 4, 5]. Andaxberg [1] gives the verloue stepo in aluster analy al a starting from the chai ce of data pointa to interpreting the results. Harti gan 5] desoribes veari oun cluatering algo- ri thms in hin book. He also states the ueos of tho ae methoda in vari oue fielde. Jardines end Sibaon [6] dovelopa meaoures of di ssimilarity and regerde a alustor mothod as a funotion from di andmilori tymatrlces to trees. Clustering tochniquos can ba broadly di vi ded into tao ootogorioe, (1) Hiererchical (11) Non hiererchicalC0. . Hiororchical olustoring techniquoa begin uith a trd angular similari ty natrix, uhoao rowa and columne corroopond to patterns and uho Be ontries moaaure almilari ty, the larger the ontries the nora similer tho pattarna. larger the antrian, the mora similer tho pattorns. The ou tput of an hiorarchicel aluatering progrem is a dondrogrom, uhi oh ia a troe shouing a soquenoa of nostod aluatoringa. This graphical output la tho main faaturo of auch programs ainao aovoral clustoringe ara roproaon tod on the osmo picture.Hiurarchical methode nay again bo divd ded into tuo oatogorios nanoly (1) agglomerativo and (11) di vi al vo 1. The goneral procodure for agglomorati ve alustoring on a data consi sting of n antitios ia as follows-1. Begin with n alusters each conal sting of exectly one anti ty. Let the clusters be labelled with the runbers 1 to n. 2. Saarah the almilari ty matrix for the most aimilar pair of alustere. Lat the choaen clueters be labelled p and q and lat their associeted similarity be agip 8pq;P>q. 3. Roduoe the number of cluetors by 1 through the morger of oluatare p and q. Labal the produot of the merger q and update the similari ty matzix ontrios in order to rofloct the rovi sed sind lari ti os betwoen alustor q and all other exd ating clusters. Dolota the rou and co luen of 5 (aimilari ty matrix) pertaining to cluster q.

Comments

ProQuest Collection ID: http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqm&rft_dat=xri:pqdiss:28842957

Control Number

ISILib-TH265

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

DOI

http://dspace.isical.ac.in:8080/jspui/handle/10263/2146

Download

Included in

Mathematics Commons

Share

COinS